Subscriber access provided by UNIV OF SCIENCES PHILADELPHIA
Chemistry and Biology of Aroma and Taste
A Comprehensive Evaluation of the Nonvolatile Chemistry Affecting the Sensory Bitterness Intensity of Highly Hopped Beers Christina Hahn, Scott Lafontaine, Clifford Pereira, and Tom H. Shellhammer J. Agric. Food Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jafc.7b05784 • Publication Date (Web): 10 Mar 2018 Downloaded from http://pubs.acs.org on March 11, 2018
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 36
Journal of Agricultural and Food Chemistry
1 Evaluation of the Nonvolatile Chemistry Affecting the Sensory Bitterness Intensity of Highly Hopped Beers Christina D. Hahn¹, Scott R. Lafontaine¹, Cliff B. Pereira¹, Thomas H. Shellhammer¹* ¹Oregon State University Department of Food Science & Technology 234 Wiegand Hall Corvallis, OR 97330 *Corresponding author. Phone: 1-541-737-9308. E-mail:
[email protected] ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 2 of 36
1
1
Abstract
2
The range of different nonvolatile constituents extracted from hops in highly hopped
3
beers suggests that isohumulones may not be the sole contributor to beers’ bitterness. Among
4
brewers producing hop-forward beer styles there is concern that the Bitterness Unit (BU) is no
5
longer an accurate predictor of beer bitterness. This study examined factors within the beer
6
matrix that influence sensory bitterness perception in highly hopped beers. Over 120 commercial
7
beers were evaluated using sensory and instrumental techniques. Chemical analysis consisted of
8
the BU via spectrophotometry, hop acids via HPLC, total polyphenols via spectrophotometry,
9
and alcohol content plus real extract via an Alcolyzer. Sensory analysis was conducted over two
10
studies, and the beers’ overall bitterness intensity were rated using a 0-20 scale. This study
11
identified that the BU measurement predicts sensory bitterness with a nonlinear response, and it
12
proposed an alternative approach to predicting bitterness based on isohumulones, humulinones,
13
and ethanol concentrations. The study also revealed the importance of oxidized hop acids,
14
humulinones, as a significant contributor to beer bitterness intensity.
15
Keywords: Hops, bitterness, brewing, beer, isohumulones, humulinones.
ACS Paragon Plus Environment
Page 3 of 36
Journal of Agricultural and Food Chemistry
1 16
Introduction
17
With the rise in popularity of hop-forward beer styles there has been a growing concern
18
in the beer industry surrounding the accuracy of instrumental bitterness analyses. Anecdotally,
19
some in the brewery quality assurance field feel that current methods for predicting beer
20
bitterness, such as the Bitterness Unit (BU) method, a liquid/liquid extraction of bitter
21
compounds, and the direct measurement of isohumulones via high performance liquid
22
chromatography (HPLC), often fail to accurately predict the sensory bitterness of beer of highly
23
hopped beer styles The projects presented herein were designed to test existing methodology and
24
to search for potential new approaches to instrumentally predicting sensory bitterness intensity in
25
beer.
26
Hops (Humulus lupulus) are the primary source of bitter compounds in beer, and these
27
bitter compounds can be categorized into three groups: isohumulones, oxidized hop acids, and
28
polyphenols.1 There is little question that isohumulones are the main contributor to beer
29
bitterness, however all three groups have the ability to impact both sensory bitterness intensity
30
and the BU value of a beer.2-5 How a brewer uses the hops will determine their concentration in
31
finished beer. The timing and mass of hop additions in the brewing process will influence the
32
overall aroma, bitterness, and astringency of the finished beer.6 In the production of hop-forward
33
beer styles, such as American craft Pale Ales and India Pale Ales (IPAs), brewers add
34
significantly more hop material to beer relative to traditional lager beer styles. This shifts both
35
the concentration and ratio of bitter compounds present in the finished beer from solely
36
isohumulones (as in American light lager beer) to a broad array of hop-derived compounds such
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 4 of 36
2 37
as humulones, oxidized humulones (humulinones), oxidized lupulones (hulupones), and
38
polyphenols, in addition to isohumulones.
39
Generally, early additions of hops during wort boiling impart more bitterness and less
40
aroma while the inverse is true for hops added late in wort and/or beer production.3 A traditional
41
hopping scheme for lager beer would be a single early addition in the range of 1.5 g/L at the
42
beginning of a 60-90 minute wort boil. By contrast, a hopping scheme for an American craft IPA
43
will involve the addition of hops at multiple stages in the brewing process and to the finished
44
beer itself, a technique referred to as “dry-hopping”. The Brewers Association reported that in
45
2014 the top selling American Imperial IPAs contained between 9 and 11.5 g/L of hop material,7
46
while American light lager beer contains approximately 1 g/L. Brewers looking to increase hop
47
aroma by adding large quantities of hops late in process also potentially increase the
48
concentration of oxidized hop acids and total polyphenols in the finished beer.
49
Oxidized hop acids are present in freshly baled, aged or improperly stored hops at
50
varying levels.8-10 Recent work has shown that oxidized hop acids, hulupones and humulinones,
51
are 84% and 66% as bitter as isohumulones, respectively.2 Separately, polyphenols can
52
contribute both bitterness and astringency to a wide variety of food and beverages depending on
53
their degree of polymerization and molecular weight, and have been shown to have an additive
54
effect on beer bitterness in the presence of isohumulones.5,
55
polyphenols contribute to sensory bitterness alone is not understood. A 2008 study showed that
56
at concentrations less than 300 mg/L, total polyphenols influence BU values in lager beer
ACS Paragon Plus Environment
11-12
The magnitude to which
Page 5 of 36
Journal of Agricultural and Food Chemistry
3 57
systems. For every 15-20 mg/L increase in total polyphenols, a one-unit increase in BU was
58
observed.5
59
At relatively high concentrations in beer (ranging 14 mg/L to 28 mg/L) humulones, the
60
precursors to isohumulones, have been shown not contribute to beer bitterness.13 However, at
61
these elevated concentrations they may influence the BU value. Although the BU method
62
captures these compounds in its measurement, there is no accounting for their relative abundance
63
or relative bitterness. Furthermore, beer matrix effects, such as alcohol, residual extract and pH,
64
may also influence bitter taste sensation.14-17 Residual extract and ethanol concentrations are well
65
known factors brewers manipulate to balance a beer’s bitterness. Typically, beers with a high
66
concentration of bittering compounds will have higher levels of residual extract.6 Ethanol
67
concentration may also play a role in bitterness perception, as numerous studies in wine show
68
that ethanol enhances bitter taste sensation.11, 17-19
69
To predict sensory bitterness intensity, the International Bitterness Unit, or BU method, is
70
most frequently employed. The BU method is a manual liquid-liquid extraction technique that
71
utilizes polarity to separate the bittering components of beer using a non-polar solvent (2,2, 4-
72
trimethylpentane).20 The absorbance of the extracted components in the non-polar phase is
73
measured via a spectrophotometer at 275 nm and multiplied by a factor of 50 to obtain the BU
74
value. The BU method was developed in 1955 by Rigby and Bethune with the intent of rapidly
75
measuring the concentration of isohumulones in beer prior to the advent of HPLC.21 By 1968,
76
the BU was internationally recognized as a standard method by the American Society of Brewing
77
Chemists (ASBC) and the European Brewery Convention.22 In addition to measuring
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 6 of 36
4 78
isohumulones, other ultraviolet-adsorbing compounds may influence the BU value.4,
20, 23
79
inclusion of non-isohumulone substances in the BU assay presents a challenge to brewers,
80
particularly when using this technique to evaluate beer styles that have been heavily hopped.4, 10
The
81
Advances in instrumentation now allow for the rapid and easy measurement of
82
isohumulones in beer using HPLC. The measurement of isohumulones by solid phase extraction
83
and HPLC was first published by Donley in 1992 and became an official method of the ASBC in
84
1993 (Beer-23, C).24 By 2011 the ASBC approved a direct injection method for the measurement
85
of isohumulones by HPLC that eliminated the need for solid-phase extraction (Beer-23, E).20, 25
86
Although HPLC allows for the measurement of a wide range of different hop acids, historically
87
the focus has been solely on isohumulones.
88
Regardless of the advances made in beer analyses tools, the American craft beers
89
produced today are unlike the beers for which the current methods (BU and isohumulones by
90
HPLC) were designed to measure. American craft beers are complex in their non-volatile
91
chemistry relative to light lager beer, and there is concern surrounding the BU’s ability to
92
adequately predict sensory bitterness in these new, highly hopped beer styles. This study aimed
93
to address this concern by (1) evaluating a broad array of chemical factors known to affect bitter
94
taste sensation along with sensory evaluation of bitterness using commercially-available,
95
American beers with a focus on hop-forward styles, (2) performing multiple linear regression
96
modeling to determine the significant drivers of beer bitterness, and (3) proposing a new,
97
analyte-specific, predictive model of beer sensorial bitterness intensity.
ACS Paragon Plus Environment
Page 7 of 36
Journal of Agricultural and Food Chemistry
5 98
Materials and methods
99
To gain insight into the drivers of sensory bitterness in beer, 121 unique commercial
100
beers from 42 breweries across the United States were tested for 7 factors known to impact the
101
sensory perception of bitterness in beers. Sensory analysis was conducted in two separate studies
102
with specific objectives for each. A multiple replication study focused on evaluating a smaller
103
number of beers (30) with a small (10 member) trained sensory panel. The beers’ overall
104
bitterness intensity were rated using a 0 to 20 scale. This study revealed the chemical drivers of
105
sensory bitterness, confirmed the inadequacy of total isohumulone content as a complete
106
measure of beer bitterness, and proposed an alternative, liquid chromatography approach to
107
predicting bitterness based on the combination of isohumulones, humulinones, and ethanol
108
concentrations. A separate, single replication study was used to validate the results from the
109
multiple replication study and was focused on evaluating a larger number of commercial beers
110
(104 in total, 13 of which had been evaluated in the multiple replication study) with a larger (19
111
member) trained sensory panel.
112
Reagents and standards
113
Ferric
ammonium
citrate
(green)
was
purchased
from
Fisher
Chemicals.
114
Ethylenediaminetetraacetic acid disodium salt, dihydrate (EDTA),
115
carboxymethylcellulose, and octyl alcohol were obtained from Sigma-Aldrich Chemical Co (St.
116
Louis, MO). HPLC grade methanol was obtained from VWR International, BDH analytical
117
(West Chester, PA, USA). Hydrochloric acid, 2,2,4-trimethylpentane, phosphoric acid and
118
ammonium hydroxide obtained from Avantor performance materials (Center Valley, PA).
ACS Paragon Plus Environment
medium viscosity
Journal of Agricultural and Food Chemistry
Page 8 of 36
6 119
DCHA-Iso ICS-I3 and international calibration extract ICE-3 standards were obtained from
120
ASBC. DCHA humulinone and hulupone standards were produced2 and pure standards were
121
obtained through Robert Smith at S.S. Steiner, Inc.
122
Beer analysis
123
The concentrations of hop acids in beer samples were assayed using a modified version
124
of ASBC method Beer-23E.20 The modified conditions were as follows: analysis was performed
125
on a 2.6 um EVO C-18 100 A LC column 100 x 4.6 mm (Phenomenex, Torrance, CA) at 40°C
126
measuring absorbance at 275 nm for the isohumulones and humulinones, 330 nm for the
127
hulupones, and 314 nm for the humulones. These wavelengths were chosen considering the
128
absorbance spectrum of each hop acid.2 Prior to analysis beer was degassed and 7 µL of each
129
beer sample was injected into a mobile phase consisting of 10% A (reagent water) and 90% C
130
(75% MeOH, 24% H2O, 1% H3PO4) flowing at 1.6 ml/min. Ten minutes following the injection
131
the mobile phase was switched to 100% B (100% MeOH), and at 14 minutes it was switched
132
back to 10% mobile phase and A 90% mobile phase C.
133
Total polyphenols were measured spectrophotometrically according to the EBC
134
Analytica methods (9.11).26 Bitterness units were measured according to ASBC methods of
135
analysis Beer 23A.20 Spectrophotometric analysis for total polyphenols and bitterness units were
136
carried out using a Shimadzu PharmaSpec UV-1700 spectrophotometer, Shimadzu Corporation
137
(Columbia, MD). Ethanol, residual extract, and pH were analyzed using an Anton-Paar
138
Alcolyzer with supporting pH module (Anton Paar USA, Ashland, VA).
ACS Paragon Plus Environment
Page 9 of 36
Journal of Agricultural and Food Chemistry
7 139
Sensory Methodology: Multiple replication study
140
For the multiple replication study, the sensory panel consisted of 10 members - 6 males
141
and 4 females, ranging from ages 21 to 54, with a median age of 28. Thirty (30) unique beers
142
were evaluated, 13 of which were also evaluated in the single replication study. The panel
143
included 4 individuals with previous training on bitterness intensity and 6 self-identified frequent
144
craft beer drinkers with minimal training on bitterness intensity. All panelists were over the age
145
of 21 and self-identified beer consumers.
146
Of the 30 beers evaluated, the majority were commercially categorized as pale ales, India
147
pale ales (IPA), session IPAs, and imperial or double IPAs. Beer selection began by evaluating
148
the chemistry of 60 unique beers from 17 American craft breweries. The beers were sorted based
149
on the highest and lowest values of the seven chemical factors of interest. The 30 beers selected
150
consisted of the extreme values of ethanol, residual extract, pH, isohumulones, humulinones, and
151
humulones in the larger beer set and represented the wide range of variation found in heavily
152
hopped beer styles.
153
The panel met as a group for five one-hour training sessions. Training sessions focused
154
on differentiating and identifying peak bitterness intensity in different beers. To minimize
155
sensory fatigue only 10 beers were evaluated per one-hour session with three separate sessions
156
making a full replication. The 30 beers were evaluated over the course of 12 sessions for a total
157
of 4 replications. To minimize variation in bitterness intensity among the sessions, a randomized
158
incomplete block design was used to ensure that the beers evaluated per session had
159
approximately the same variance in BU values (See Supporting Information). Each 10-beer
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 10 of 36
8 160
group was evaluated in its own session, and Qualtrics software (Qualtrics, Provo, UT) was used
161
to generate a random presentation order for each panelist within each session. Beers samples (45
162
ml) were presented in open 120 ml clear glasses, blind coded with random three-digit numbers.
163
Beers were stored at 4°C for the duration of the study and removed from cold storage and poured
164
30 minutes prior evaluation. Training and testing took place in the same location; a conference
165
room held at 20°C and lit with Fluorescent lighting.
166
At the beginning of each session, panelists were assigned a Chromebook tablet (Google,
167
Mountain View, CA), and directed to an online Qualtrics survey. Each testing session began with
168
the panelists warming up and calibrating bitterness intensity scaling by tasting four reference
169
samples. External reference samples for anchoring the overall bitterness scale were chosen by
170
using four commercially available beers of varying sensory bitterness intensities. The sensory
171
scores for these references were determined by consensus during training. The zero reference
172
point (0) was anchored by Michelob Ultra-Light (Anheuser-Busch, St. Louis, MO) (BU = 4), the
173
low reference point (4) was anchored by Session Lager (Full Sail Brewing Co., Hood River, OR)
174
(BU = 18)., the medium-low reference point (8) was anchored by Mirror Pond Pale Ale
175
(Deschutes Brewery, Bend, OR) (BU = 40), and the medium-high reference point (15) was
176
anchored by Torpedo IPA (Sierra Nevada Brewing Co., Chico, CA) (BU = 66). No reference
177
was selected to anchor the high end of the scale (20) thereby reducing potential sensory fatigue.
178
After the reference samples were evaluated, panelists were instructed to taste the samples
179
in the order in which they appeared on their ballot. Panelists scaled the overall bitterness
180
intensity of a given sample using a sliding bar scale (0 = none, 20 = extreme), and the scaling
ACS Paragon Plus Environment
Page 11 of 36
Journal of Agricultural and Food Chemistry
9 181
data were collected using Qualtrics. After assessing a sample, the panelists were instructed to
182
rinse their mouths with a 0.1% pectin solution to aid in the reduction of bitterness carry-over,35
183
and then to rinse their mouths again with deionized water. Panelists were instructed to wait until
184
all perceivable bitterness had dissipated before tasting the next sample. Panelists repeated this
185
procedure for each sample. After all samples had been assessed, panelists were allowed to re-
186
taste the samples (if necessary) in any order deemed necessary to most accurately assess a given
187
sample’s overall sensory bitterness intensity.
188
Sensory Methodology: Single replication study
189
For the single replication study, the panel consisted of 19 members - 14 males and 5
190
females, ranging from ages 22 to 65, with a median age of 28. The panel contained 4 individuals
191
with previous training on bitterness intensity and 15 self-identified frequent craft beer drinkers
192
with minimal training on bitterness intensity. Of the 19 panelists, 4 also participated in the
193
multiple replication study. The study started with 116 unique beers from 42 American breweries
194
but was reduced to 104 unique beers from 38 American breweries, due to quality defects
195
(preliminary data not shown). Thirteen (13) of the 104 beers had also been evaluated in the
196
multiple replication study. This study included a greater number of beer styles in an effort to
197
capture the variability within styles produced by craft and macro breweries alike. The beers were
198
not screened based on their chemical composition prior to sensory evaluation.
199
The panel met as a group for three one-hour training sessions. Training was followed by
200
twelve testing sessions conducted over the course of ten weeks. Panelists evaluated 10 beers per
201
session and scaled bitter intensity using a paper ballot. Paper ballots were utilized due to the
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 12 of 36
10 202
limited number of available Chromebooks and the increased number of panelists. XLSTAT
203
software was used to randomly assign the beers to each session, and to randomize the sample
204
presentation order for each panelist. Two beers were selected to be replicated throughout the
205
study thereby serving as internal controls: Fat Tire (New Belgium Brewing, Fort Collins, CO)
206
served as a low bitterness intensity control (overall bitterness intensity = 3) (BU = 20)and
207
Torpedo IPA (Sierra Nevada Brewing Co., Chico, CA) (BU = 66) served as the medium-high
208
bitterness intensity control (overall bitterness intensity = 15). These beers were randomly
209
replicated three times during the study. The remainder of the samples were not replicated.
210
Panelists scaled the overall bitterness intensity of a given sample using a sliding bar scale (0 -
211
20) on a paper ballot. Sensory training, commercial reference beers, environmental conditions,
212
and general sensory protocol was carried out as described previously for the multiple replication
213
study.
214
Statistical analysis
215
In the multiple replication study, each of the 30 beers was evaluated 40 times,
216
independently (10 panelists over 4 replications). Panel performance ANOVA revealed
217
significant main effects for panelist, beer, and panelist by beer interaction (Table 1). A
218
significant panelist effect is not unexpected in sensory analysis as individual panelists interpret
219
stimuli differently. No effects were observed for replication, replication by panelist, or
220
replication by beer. The 40 bitterness intensity scores generated for each beer were averaged into
221
a single value and assigned to that beer as an overall bitterness intensity score. Results were
222
evaluated using a mixed model, two-way, analysis of variance (ANOVA) model. Fixed effects
ACS Paragon Plus Environment
Page 13 of 36
Journal of Agricultural and Food Chemistry
11 223
included beer. Random effects included panelist, replication, panelist by beer, replication by
224
beer, and panelist by replication interactions. Data analysis was carried out using XLStat 2016
225
(Addinsoft, New York, NY, U.S.A.).
226
In the single replication study, most panelists missed at least one session during the study,
227
thus each of the 104 unique beers was evaluated singly by 9 to 19 panelists. This unbalancedness
228
with respect to number of panelists per beer required the generation of adjusted means for each
229
beer (adjusted for panelist differences) to be used as the Overall Bitterness Intensity score for
230
each beer.
231
Multiple linear regression using stepwise (forward) model selection in the SAS
232
GLMSELECT procedure was carried out for the seven predictive chemical factors of interest
233
(linear and quadratic in each factor as well as linear-by-linear interactions) to develop a model to
234
predict the overall bitterness intensity scores that were derived from each of the two sensory
235
panel studies. The model selection criterion used was AICC (Corrected Akaike’s Information
236
Criterion) and it was found that the same conclusions would be reached with other criteria
237
(Akaike’s Information Criterion (AIC), Schwartz’s Bayesian Criterion (SBC) or prediction sum
238
of squares (PRESS)). A model hierarchy requirement was used so that quadratic in a predictor
239
could enter the model only when linear in that predictor was already present in the model and a
240
linear-by-linear interaction could enter the model only when the two linear terms were already
241
present in the model. Resampling and model selection for each resampled set of data was used
242
to assist in determining the relative importance of predictors (effect selection percentage).
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 14 of 36
12 243
Second order polynomial models and other linear models were fit with the SAS
244
GLMSELECT procedure. Nested models were compared using F tests (numerator = the change
245
in model sum of squares divided by the change in number of parameters; denominator = residual
246
mean square from the full model).
247
Results and Discussion
248
The beers evaluated in this study presented a broad range of chemistry and sensory
249
bitterness (Figure 1). Furthermore, the beer subsets as a group for each study were not
250
statistically different as assessed by a means comparison (2-sided t-test, p>0.05).
251
Across both studies, residual extract concentrations ranged from 2.5 to 8.1% w/w with an
252
average concentration of 4.9% w/w (Figure 1.A). Residual extract concentration is largely
253
determined by beer style. Lager beers often have lower residual extract relative to ales such as
254
porters, stouts, pale ales and IPAs. Residual extract is mostly comprised of the un-fermentable
255
carbohydrates that remain following the alcoholic fermentation, and these dextrins are primarily
256
lower molecular weight polymers of D-glucose units linked by α-(1→4) or α-(1→6) glycosidic
257
bonds. The breakdown of these dextrins in the oral cavity can lead to the sweet taste of some
258
beers.14-15,
259
extract to influence the flavor, mouthfeel, and bitterness of beer. Beers with increased bitterness
260
tend to be brewed with a higher residual extract, a technique used by brewers to create
261
“balanced” beers.6 Across all food systems, sweet taste suppresses bitter taste, and bitter taste
262
suppresses sweet taste.14,
263
taste qualities exhibit this inhibitory pattern.29-30
27
The residual extract also contains protein and minerals. Brewers adjust residual
28
This phenomenon is known as mixture suppression, and all basic
ACS Paragon Plus Environment
Page 15 of 36
Journal of Agricultural and Food Chemistry
13 264
Like residual extract, the ethanol concentration of a beer may be determined by beer
265
style, however significant variation in alcohol concentration may also be found within styles. It is
266
not uncommon to find “sessionable” (low ethanol) or “imperial” (high ethanol) versions of a
267
brewery’s flagship beer seasonally. The beers in this study displayed a wide range of ethanol
268
concentrations from 3.8% to 10.1% v/v. The average concentration of ethanol across both studies
269
was 6.4% v/v (Figure 1.B). The influence of ethanol on beer flavor is multifaceted, and its effect
270
on bitterness intensity is one example. Ethanol is considered to have a sweet taste at low
271
concentrations,18 however numerous studies have shown that the presence of ethanol increases
272
bitter taste sensation in alcoholic beverages.11,
273
evaluating bitterness intensity of commercial beer, as many beers tend increase in both residual
274
extract and ethanol concentration as hopping rates increase.
16-19, 28
This can be a confounding factor when
275
The range of pH values was somewhat wider than expected at pH 4.0 to 5.4 (Figure 1.C).
276
The elevated pH values are likely the result of beers with high levels of dry-hopping.8 It is worth
277
noting, that as pH increases, so does the solubility of humulones (α-acids), lupulones (β-acids),
278
and humulinic acid. A 1955 study by Spetsig showed that the solubility of these hop acids
279
increased exponentially with pH.31 Thus, the effect of elevated pH via dry hopping may increase
280
the concentration of humulones in finished beer.
281
The solubility of humulones in wort and beer is low, and thus it was not unexpected to
282
find very low levels of these compounds in the beers in this study. While the solubility of
283
humulones is pH dependent, it has been reported to be approximately 3 mg/L in water.32 In a
284
traditional brewing process, where the hop material is boiled in wort for an extended period of
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 16 of 36
14 285
time (60-90 minutes), very little humulone will persist into the finished beer.1 In dry-hopped beer
286
styles, the solvating power of ethanol increases the solubility of humulones. The majority of the
287
beers tested in the study presented herein had little to no humulones, however four of the beers
288
had humulone concentrations greater than their solubility in beer (14 mg/L)13 with a maximum of
289
24.2 mg/L (Figure 1.F). Although humulones did not contribute bitterness to beer at these levels,
290
they do adsorb at 275 nm and may contribute to a beer’s BU value if present. Because
291
humulones may contribute to the BU value, but are not bitter, they may affect the BU’s ability to
292
accurately predict bitterness intensity.13, 23
293
As previously stated, isohumulones are the primary source of bitter taste in beer. They are
294
isomers of hop-derived humulones which are transformed with heat and time via a first order
295
reaction that is temperature-dependent.3 Isohumulone concentrations (as determined by HPLC)
296
averaged 36 mg/L across both studies and ranged from 7.0 to 74.2 mg/L (Figure 1.D). The most
297
prominent forms of isohumulone are isohumulone, iso-co-humulone and iso-ad-humulone,33-35
298
and it is standard practice to sum their concentrations to obtain the reported value of
299
isohumulone.20
300
The role of oxidized hop acids has been the focus of several studies related to hop and
301
beer quality.2, 36-38 Although there is little evidence to support the theory, there is a widespread
302
belief within the brewing community that as hops age and decrease in total α-acids the increase
303
in oxidized lupulones (hulupones) prevents the loss of bittering potential in the hop itself.32 Of
304
the 121 unique beers tested in this study, most had no detectible levels of lupulones or their
ACS Paragon Plus Environment
Page 17 of 36
Journal of Agricultural and Food Chemistry
15 305
oxidation products. In a small number of cases, very low levels of hulupones were detected, but
306
these levels were below the limit of quantification.
307
In contrast, both the prevalence and concentration of the oxidized humulones
308
(humulinones) (Figure 1.E) were higher than previously reported in literature. The average
309
concentration of humulinones was 17 mg/L and 117 out of the 121 unique beers contained
310
detectible levels of humulinones (≥1 mg/L). The concentration of humulinones in finished beer
311
is dependent on the concentration of humulinones present in the hop varieties used, total hop
312
mass, and hop contact time.4, 8 A recent study by Algazzali and Shellhammer confirmed the bitter
313
quality of humulinones as previously published.35,
314
detect humulinones as low as 8 mg/L, and it was proposed that humulinones have a relative
315
bitterness intensity of 66% when compared to isohumulones.2 A recent study by Oladokun et
316
al.40 examined the impact of hop bitter acids and polyphenols on beer bitterness in 34 global
317
commercial lager beers. In this study humulinone concentrations were considerably lower (≤ 3
318
mg/L) in the lager beers tested in comparison to the American hoppy ales and lagers tested in
319
this study.40
39
The study’s trained sensory panel could
320
The concentration of total polyphenols in finished beer is dependent on hop mass, hop
321
type (extract, pellet or whole cone), the hop varieties used, the temperature of dry-hopping and
322
contact time.9-10 Polyphenols comprise a vast group of different compounds that contribute to
323
bitterness and/or astringency of beer.41 Typically, heavily late-hopped and/or dry-hopped beer
324
styles contain elevated levels of total polyphenols.40 A study by McLaughlin and Shellhammer5
325
showed polyphenols to have an additive effect on bitter taste perception when combined with 10
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
Page 18 of 36
16 326
mg/L isohumulones in lager beer. The magnitude in which polyphenols alone influence the
327
bitterness of beer is currently not understood. However, Oladokun et al.40 showed polyphenols to
328
influence the bitter quality of lager beer. In the study, beers with increased concentrations of
329
polyphenols were perceived as having ‘harsh’ and ‘progressive’ bitterness. It is worth noting that
330
these studies were conducted in lager beers with maximum total polyphenol concentrations
331
below 300 mg/L. In this study, total polyphenol concentrations ranged from 135 to 697 mg/L
332
with an average of 329 mg/L (Figure 1.G). Over 60% of the beers tested in the study presented
333
herein contained total polyphenol concentrations in excess of 300 mg/L. Additional research is
334
needed to understand the impact of polyphenols in beer at these concentrations.
335
The BU method was developed in 1955 as a way to rapidly measure isohumulones prior
336
to the advent of the HPLC.21 When analyzing beers in which isohumulones are the primary
337
source of bitter compounds, the BU is an accurate predictor of sensory bitterness. In hop-forward
338
beer styles, such as American craft IPAs and pale ales, brewers add hops to the wort late in
339
production or to finished beer. This technique preserves volatile hop aroma compounds that
340
would otherwise be lost during wort boiling. In addition to adding hop aroma, late additions of
341
hops also add non-volatile compounds such as humulones, oxidized hop acids, and
342
polyphenols.42 These compounds may contribute to a beer’s BU value, and at elevated
343
concentrations may potentially impact a beer’s bitterness intensity. In this study, bitterness units
344
ranged from 10 to 118 with an average BU of 60 (Figure 1.H). A mistaken conventional wisdom
345
is that 1 mg/L of isohumulones is equal to 1 BU. Although this may hold true for low bitterness
346
lager beers, the relationship did not hold true for the beers evaluated in this study (Figure 2).
ACS Paragon Plus Environment
Page 19 of 36
Journal of Agricultural and Food Chemistry
17 347
The beers tested had a wide range of sensory bitterness intensities (Figure 1.I). A variety
348
of breweries, brands, and beer styles where tested, but an emphasis was placed on hop-forward
349
beer styles, as one of objectives the study was to understand if the BU could accurately predict
350
sensory bitterness intensity in these beers. At low levels of BU, the relationship between sensory
351
bitterness intensity and the BU appeared to be linear. However, for BU values of 50 and above,
352
the relationship became curvilinear, and increases in sensory bitterness intensity became smaller
353
as the BU values increased (Figure 3, D & 4 D). This result may be from a saturation
354
phenomenon, that is, as stimuli increases, it becomes increasingly difficult to differentiate
355
changes in the stimuli as the senses become overwhelmed.30 Alternately, it may be that the
356
compounds affecting increases in the BU at these elevated levels (> 50 BU) have a reduced
357
impact on sensory bitterness intensity, or some combination of the two.
358
For each of the two studies, model selection for predicting sensory bitterness intensity
359
was conducted using all seven factors of interest (linear, quadratic and linear-by-linear
360
interactions). For both studies, the most important predictors were linear in isohumulones, linear
361
in humulinones, linear in alcohol by volume (ABV) and second-order terms involving
362
isohumulones and/or humulinones). Because of the importance of linear and second-order effects
363
involving isohumulones and humulinones, the second order response surface was investigated by
364
fitting a full second-order model to the data (Eqn. 1).
365
BI = β₀ + α₁(ISO) + γ(HU) + δ(ISO * HU) + α₂(ISO)² + γ₂(HU)²
366
where BI = sensory bitterness intensity (0-20),
367
ISO = Isohumulones and HU = humulinones (mg/L)
ACS Paragon Plus Environment
(Eqn. 1)
Journal of Agricultural and Food Chemistry
Page 20 of 36
18 368
Within each study, the fitted model had the following characteristics:
369
1) linear coefficients were similar to each other (α1 ≅ γ1);
370
2) quadratic coefficients were similar to each other (i.e. α2 ≅ γ2);
371
3) the interaction coefficient (δ) had the same sign as the quadratic terms (α2 & γ2) and
372
(given the large standard error) was not very different from the sum of the quadratic
373
coefficients.
374
These three characteristics (and contour plots of the fitted surface) suggested that a much simpler
375
model with a single predictor (sum(ISO + HU) = the sum of the isohumulones and humulinones)
376
could predict sensory bitterness intensity well (Eqn. 2). Indeed, in both studies the simpler model
377
fit well (R2= 0.913 in the multiple rep study and R2 = 0.772 in the single-rep study) and the more
378
complex model (Eqn. 1) did not significantly improve the fit of the simpler model (F < 1, p >
379
0.49 for both studies).
380
= + + + +
381
The effectiveness of sum(ISO + HU) as a predictor is shown graphically in Figures 3 and
382
4. Individually, isohumulone and humulinone concentrations are relatively poor predictors of
383
sensory bitterness (Figures 3 & 4, A & B), while the sum of isohumulones and humulinones
384
(Figure 3 & 4, C) is almost as good a predictor of sensory bitterness as BU (Figure 3 & 4, D).
385
These same figures also highlight a need for the model to capture curvilinearity as well as the
386
greater variation about the relationship in the single-rep study (Figure 4) compared to the
387
multiple rep study (Figure 3).
ACS Paragon Plus Environment
(Eqn. 2)
Page 21 of 36
Journal of Agricultural and Food Chemistry
19 388
With the effectiveness of a single predictor established, model selection was repeated
389
with sum(ISO + HU) as an 8th potential predictor. In both studies, the most important predictors
390
were linear and quadratic in sum(ISO + HU) and linear in ABV (Eqn. 3). The addition of ABV
391
increased the R-squared slightly, but significantly (p = 0.0002 for single-rep study and p=0.041
392
in multiple rep study). The positive influence of ethanol on bitterness perception is not
393
unexpected as it has been reported elsewhere that ethanol increases bitter perception in beer and
394
wine.16-19, 27-28
395
= + + + + + ₃
396
It is important to note that although the simple sum of isohumulones and humulinones
397
was effective as a single predictor, this does not mean that future studies should necessarily give
398
equal-weight to isohumulones and humulinones. That is, a weighted sum of isohumulones and
399
humulinones could potentially be even more effective as a predictor in future studies despite not
400
being the case in this study.
(Eqn. 3)
401
As it was devised, the BU describes the bitterness of beer in a linear fashion, and for low
402
BU beers this relationship holds true. However, when using a linear BU response to predict
403
bitterness intensity over the entire data set within each study, there was obvious systematic lack
404
of fit with the tendency to under-predict in the mid-range of bitterness and over-predict at the
405
extremes. Adding a quadratic term significantly improved the fit (p F Beer 29 11826 407.8 32.98