Pain Medicine, 0(0), 2019, 1–11

doi: 10.1093/pm/pnz022
Preliminary Research Article

Downloaded from https://academic.oup.com/painmedicine/advance-article-abstract/doi/10.1093/pm/pnz022/5380130 by guest on 14 March 2019

The Effect of Low-Carbohydrate and Low-Fat Diets on Pain in
Individuals with Knee Osteoarthritis
Larissa J. Strath,* Catherine D. Jones, MSPH,* Alan Philip George,* Shannon L. Lukens,* Shannon A.
Morrison, PhD, CRNP,† Taraneh Soleymani, MD,‡ Julie L. Locher, PhD,§ Barbara A. Gower, PhD,¶ and
Robert E. Sorge, PhD*
*Department of Psychology; †School of Nursing; ‡Department of Nutrition Sciences; §Division of Gerontology, Geriatrics and Palliative Care,
Department of Medicine; and ¶Department of Nutrition Sciences, University of Alabama at Birmingham, Birmingham, Alabama, USA

Funding sources: This work was supported in part by a Translational Nutrition and Aging Research Academic Career leadership Award (K07AG043588-
03) to JLL. RES contributed startup funds.

Conflicts of interest: The authors declare no conflicts of interest.

Correspondence to: Robert E. Sorge, PhD, University of Alabama at Birmingham, 1300 University Blvd., CH 415, Birmingham, AL 35294, USA.
Tel: 205-934-8563; Fax: 205-975-6110; E-mail: rsorge@uab.edu.

Abstract
Objective. Osteoarthritis is the most prominent form of arthritis, affecting approximately 15% of the population in the
United States. Knee osteoarthritis (KOA) has become one of the leading causes of disability in older adults. Besides
knee replacement, there are no curative treatments for KOA, so persistent pain is commonly treated with opioids,
acetaminophen, and nonsteroidal anti-inflammatory drugs. However, these drugs have many unpleasant side
effects, so there is a need for alternative forms of pain management. We sought to test the efficacy of a dietary inter-
vention to reduce KOA. Design. A randomized controlled pilot study to test the efficacy of two dietary interventions.
Subjects. Adults 65–75 years of age with KOA. Methods. Participants were asked to follow one of two dietary interven-
tions (low-carbohydrate [LCD], low-fat [LFD]) or continue to eat as usual (control [CTRL]) over 12 weeks. Functional
pain, self-reported pain, quality of life, and depression were assessed every three weeks. Serum from before and af-
ter the diet intervention was analyzed for oxidative stress. Results. Over a period of 12 weeks, the LCD reduced pain
intensity and unpleasantness in some functional pain tasks, as well as self-reported pain, compared with the LFD
and CTRL. The LCD also significantly reduced oxidative stress and the adipokine leptin compared with the LFD and
CTRL. Reduction in oxidative stress was related to reduced functional pain. Conclusions. We present evidence sug-
gesting that oxidative stress may be related to functional pain, and lowering it through our LCD intervention could
provide relief from pain and be an opioid alternative.

Key Words: Diet; Knee Pain; Osteoarthritis; Oxidative Stress; Intervention; Leptin

Introduction United States due to the increased presence of an aging,

Osteoarthritis (OA) is the most predominant form of ar-
thritis, affecting about 15% of the population [1].
Because of its affinity for the lower, weight-bearing
joints, knee osteoarthritis (KOA) has become one of the
leading causes of disability in the lower extremities in an
ever-aging population, with a lifetime risk estimated at
40% in men and 47% in women [2]. The risks are even
higher in those who are classified as obese, and the preva-
lence is expected to double from 2014 to 2020 in the
obese population [3]. Thus, treatments must be devel-
oped that can address both the obesity and resulting pain
due to OA.
Risk factors for developing KOA, aside from age and
obesity, include major mechanical injury, repetitive use
and abnormal loading of the joints, sex, race/ethnicity,
diet, and other genetic factors [4]. Currently there are no
curative treatments for KOA besides total replacement of
the joint, and the symptomatic pain that persists in this

C 2019 American Academy of Pain Medicine. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
V 1
2 Strath et al.
population is commonly treated with opioid medications,
acetaminophen, and nonsteroidal anti-inflammatory
drugs. These treatments often reduce pain but come with
the cost of substantial adverse side effects [5]. As there is
a lack of balance between pain relief and side effects,
there needs to be a solution to the experienced pain of
KOA without the downfalls of medication.
The damage to the cartilage and bone in KOA
prompts an immune response that leads to inflammation
and subsequent pain within the joint [6]. Specifically, lev-
els of C-reactive protein (CRP) and inflammatory cyto-
kines such as interleukin-6 (IL-6), IL-1b, IL-18, and
tumor necrosis factor alpha (TNF-a) were elevated in
individuals with KOA [7,8]. In addition, anti-
inflammatory mediators such as IL-10 were shown to be
decreased in those with KOA [8]. Given that inflamma-
tion plays a key role in OA patients, it is also important
to consider how diet may play a role in inflammation and
subsequent pain. Typical American diets are high in re-
fined carbohydrates and saturated fats [9]. Previous stud-
ies have shown that high levels of carbohydrates
consumed in the diet can lead to the formation of ad-
vanced glycation end products (AGEs), a pro-
inflammatory process that appears to increase the pro-
duction of pro-inflammatory cytokines [10]. IL-6 and
TNF-a can induce the expression of pro-inflammatory
mediators and degradative enzymes that inhibit cartilage
matrix synthesis and slow down the already subdued car-
tilage repair process [8]. In addition, increased CRP levels
are a risk factor for chronic pain and are also associated
with higher pain sensitivity [11]. Our previous work in
rodents has shown that a diet high in carbohydrates and
fats increases microglial activation in the spinal cord,
increases peripheral inflammation, and prolongs hyper-
sensitivity after inflammatory injury in rats [12] and mice
[13,14]. In addition, we have shown that a low-glycemic
diet leads to more rapid recovery and lower pain hyper-
sensitivity after the same inflammatory injury [13]. It is
possible that the high levels of carbohydrates and un-
healthy fats in the American diet could be contributing to
the inflammation in OA and therefore worsening the
pain that patients are experiencing. By lowering the in-
take of refined carbohydrates, we speculate that 1) the
amount of thiobarbituric acid reactive substances
(TBARS; a measure of oxidative stress as a product of
AGEs) will decrease and 2) a reduction in dietary carbo-
hydrates will improve functional pain. Diets such as the
Mediterranean diet (a partial low-carbohydrate diet)
have been shown to reduce inflammation in arthritis
patients [15] and self-reported pain in OA [16] and rheu-
matoid arthritis [17,18], supporting our hypothesis.
However, diet intervention studies to date have focused
exclusively on self-reported pain [16–20], and not assess-
ment of functional pain, which may be a better indicator
of efficacy. This randomized controlled pilot study aims
to provide support for the use of a low-carbohydrate diet
to reduce pain related to KOA, irrespective of weight
loss, as a potential alternative to pharmacological
therapies.
Methods
Participants
Twenty-one adults (nine males, 12 females) between the
Downloaded from https://academic.oup.com/painmedicine/advance-article-abstract/doi/10.1093/pm/pnz022/5380130 by guest on 14 March 2019
ages of 65 and 75 years with medically diagnosed knee
osteoarthritis (based on the American College of
Rheumatology criteria [21]) were recruited via telephone
from research databases or through community adver-
tisements. Inclusion criteria also included the ability to
understand verbal or written English and willingness to
follow a prescribed diet. Exclusion criteria included
preexisting metabolic conditions (e.g., diabetes), inability
to read/speak English, unwillingness to follow a pre-
scribed diet, currently on a prescribed diet, history of
joint replacement, history of eating disorders, difficulty
chewing or swallowing, current or recent smoker (last six
months), dependence on others for food preparation, car-
diovascular disease or renal failure (last six months), re-
cent significant weight loss (>4 kg in last two months),
uncontrolled blood pressure (SBP >159 or DBP
>95 mmHg), daily opioids for pain, and other medica-
tions that may interfere with study outcomes (e.g., pro-
ton pump inhibitors). All participants provided written
informed consent, as approved by the University of
Alabama at Birmingham Institutional Review Board.
Dietary Interventions
Following determination of eligibility, participants were
randomly assigned to one of three diet groups (Table 1):
low-carbohydrate (LCD; N ¼ 8), low-fat (LFD; N ¼ 6),
and control (CTRL; N ¼ 7). The LCD group was
instructed to restrict their daily total (not net) carbohy-
drate intake to 20 g/d for the first three weeks and was
further permitted to increase daily intake to 40 g, if re-
quired. Fats were not restricted, nor were protein (meats,
eggs). Fruits were restricted, and vegetables were permit-
ted in limited quantities (two cups per day of leafy
greens, one cup per day of nonstarchy vegetables, etc.).
Participants were instructed as to the types and quantities
of beverages that were permitted to accompany the LCD.
Artificial carbohydrate-free sweeteners (stevia or sucra-
lose) were permitted, but powdered sweeteners (aspar-
tame, saccharin, stevia, sucralose) were allowed in
limited quantities as they contain maltodextrin (1 g of
rapidly digesting carbohydrate). The LFD group was
used as a weight loss control, in which males (500 kcal/d)
and females (250–300 kcal/d) had their daily calories re-
duced primarily through reduction of fats, according to
the United States Department of Agriculture guidelines
(www.choosemyplate.gov). This diet results in approxi-
mately 60% of daily kcal from carbohydrates, 20% from
protein, and 20% from fats. The LFD consists of fruits
and vegetables, low-fat foods, whole grains, low-fat
Diet Interventions for Knee Pain 3

Table 1. Macronutrient overview of the three different diet groups

Diet Calories/Day Fat, g/d Carbohydrate, g/d Protein, g/d

Low-fat diet 800–1,200 50–67 Unlimited 100
Low-carbohydrate diet Unlimited Unlimited 20 100
Control diet Unlimited Unlimited Unlimited Unlimited

Downloaded from https://academic.oup.com/painmedicine/advance-article-abstract/doi/10.1093/pm/pnz022/5380130 by guest on 14 March 2019

Table 2. Testing parameters, frequency, and references for each outcome measure

Measure Type of Test Frequency Reference

Body weight, kg Anatomical Every 3 wk N/A
Body mass index Anatomical Every 3 wk N/A
Fat mass Anatomical Every 3 wk N/A
Food journal Adherence Daily N/A
Temporal summation Functional Every 3 wk [21]
Timed walk Functional Every 3 wk [21]
Repeated chair stands Functional Every 3 wk [21]
Leptin Blood/serum Every 6 wk N/A
IFN-c Blood/serum Every 6 wk N/A
IL-6 Blood/serum Every 6 wk N/A
TNF-a Blood/serum Every 6 wk N/A
CRP Blood/serum Every 6 wk N/A
TBARS Blood/serum Every 6 wk N/A
Knee Injury and Osteoarthritis Outcome Score Self-report/behavioral Every 3 wk [22]
Brief Pain Inventory Self-report/behavioral Every 3 wk [23]
Quick Inventory of Depressive Symptomology Behavioral Every 3 wk [24]

CRP ¼ C-reactive protein; IFN ¼ interferon; IL ¼ interleukin; TBARS ¼ thiobarbituric acid reactive substances; TNF ¼ tumor necrosis factor.

dairy, and limited cholesterol and saturated fats. The
CTRL group was permitted to eat as usual, though they
were given educational documents related to portion
control.
Outcome Measures
A summary of all measures and their timing is shown in
Table 2.
Body weight (kg), body mass index (BMI), and fat
mass (FM) were obtained using a body composition ana-
lyzer, TBF-310 (Tanita, Tokyo, Japan). Waist circumfer-
ence (cm) and height (cm) were obtained using a
measuring tape at each visit. Participants were given a
food journal to log their meals each week. The type of
food, beverages, sauces, and added sugars/sweeteners
were recorded, as well as the amounts consumed.
Adherence was verbally confirmed, and food journals
were assessed by a Registered Dietician and the study ad-
ministrator at each visit to ensure compliance with the
assigned diet. In the event of nonadherence, participants
were counseled with respect to the assigned diet.
Persistent nonadherence (more than seven days with ex-
cess carbohydrates or fats) resulted in removal from the
study. At each visit, a physician assessed their overall
physical health (blood pressure, heart rate, temperature,
weight loss/gain, adverse symptoms) to ensure the
diet was not compromising health. Participants returned
every three weeks for testing, for a total of
12 weeks (T0 ¼ baseline, T1 ¼ 3 weeks, T2 ¼ 6 weeks,
T3 ¼ 9 weeks, T4 ¼ 12 weeks). All data were collected at
the Susan Mott Webb Building for Nutrition Sciences at
the University of Alabama at Birmingham.
Functional Pain Testing
The participants rated the intensity/unpleasantness of
any pain in the affected knee on a scale from 0 to 100
(0 ¼ no pain, 100 ¼ the worst pain imaginable) before
and after participating in the functional pain testing
tasks. Functional pain scores were defined as the differ-
ence in pain scores (intensity and unpleasantness) given
before and after completing the tasks. Instructions for
rating scales were played on each visit on an Android
tablet to ensure consistency of instruction. For each task,
instructions and the description of the task were played
on the tablet before each test. Participants were given a
five-minute break between the tasks to allow for recov-
ery. The repeated chair stands and timed walk are
components of the Short Physical Performance Battery,
and all procedures were followed as described (see
below) [22].
Temporal Summation
Participants were asked to sit in a chair with both feet
flat on the floor and both knees bent at approximately
90 . Participants were asked to rate the intensity and un-
pleasantness of the pain in the affected knee separately
on a scale from 0 to 100 before beginning the test. A
300-g von Frey filament was presented once on the cen-
tral anterior patella of the affected knee for one second.
After one presentation, participants again rated intensity
4 Strath et al.
and unpleasantness. Ten presentations of the filament
were then applied to the same area of the knee (1/sec).
After 10 presentations, participants again rated the inten-
sity and unpleasantness. The increase following the single
presentation and the repeated presentation was taken as
a measure of temporal summation.
Timed Walk
A distance of eight feet was marked out in a straight line
on the floor of the testing room with two pieces of tape.
Before the walk, participants were asked to rate the in-
tensity and unpleasantness of the pain in the affected
knee while standing. Participants were asked to stand at
one end and walk to the other end (tape to tape) and
stop. The walk was timed (ms) using a handheld stop-
watch from the moment the participant initiated move-
ment to the moment they crossed the second piece of tape
(eight feet). The participant completed the timed walk
twice. The mean of the two the times was assigned to the
participant as an ordinal score (0 ¼ could not do, 1 ¼
>5.7 sec [<0.43 m/sec], 2 ¼ 4.1–6.5 sec [0.44–0.60 m/
sec], 3 ¼ 3.2–4.0 sec [0.61–0.77 m/sec], 4 ¼ <3.1 sec
[>0.78 m/sec]). If the task was started but not completed,
an ordinal score of 0 was assigned.
Repeated Chair Stands
Participants were asked to sit in a chair with their knees
bent and with feet flat on the floor. Before starting the
test, participants rated the intensity and unpleasantness
of the pain in their affected knee. Participants were asked
to place their hands on the opposite shoulders (right
hand/left shoulder, left hand/right shoulder) and then
progress from sitting to standing and back a total of five
times without using their arms. The test was timed from
the moment the participant initiated movement until they
sat down for a fifth time. Ordinal scores were assigned
(0 ¼ could not do, 1 ¼ >16.7 sec, 2 ¼ 16.6–13.7 sec,
3 ¼ 136.6–11.2 sec, 4 ¼ <11.1 sec) based on how long it
took the individual took to complete all five chair stands.
If the task was too difficult and/or not completed, the
participants were allowed to stop and given an ordinal
score of 0, and the number of chair stands completed was
recorded.
Blood Draw and Analysis Methods
Blood was collected at the first visit and at six and
12 weeks by a Registered Nurse. Tiger Top tubes (Fisher
Scientific, 8.5 mL) with clot-activating gel were used to
collect the blood and separate serum from the cells.
Samples were left at room temperature for 15 minutes af-
ter blood draw. The tubes were then centrifuged at
15,000 rpm at 4 C for 15 minutes. The supernatant was
then pipetted into two 1-mL screw-top lid tubes, labeled
and stored at –80 C until analysis. Analyses were per-
formed at the Diabetes Research Center Human
Physiology Core Laboratory (P30DK056336) at The
Table 3. Demographic data for participant sample
Variable CTRL LFD LCD P Value
Sex (M/F) 3/4 3/3 3/5 0.896
Race 2/4/1 5/1/0 7/1/0 0.120
(NHW/AA/HSP)
Age, y 68.71 6 7.11 72.33 6 1.97 71.00 6 3.12 0.230
No. randomized 4/5 3/4 3/5 N/A
Downloaded from https://academic.oup.com/painmedicine/advance-article-abstract/doi/10.1093/pm/pnz022/5380130 by guest on 14 March 2019
(M/F)
AA ¼ African American; CTRL ¼ control; F ¼ female; LCD ¼ low-carbo-
hydrate diet; LFD ¼ low-fat diet; M ¼ male; NHW ¼ non-Hispanic white;
HSP ¼ Hispanic.
University of Alabama at Birmingham. CRP (mg/L) was
analyzed on the Stanbio Sirrus Clinical Chemistry
Analyzer (Thermo Fisher Scientific, Pittsburgh, PA, USA)
using turbidometric reagent (Pointe Scientific, Canton,
MI, USA). Leptin (ng/mL) was analyzed using the
Millipore RIA Assay kit (Millipore Sigma/Merck KGaA,
Darmstadt, Germany), with a minimum sensitivity (90%
bound) of 1.269 ng/mL and an intra-assay CV of
7.044%. IFN-c, IL-6, and TNF-a (pg/mL) were analyzed
using the MSD Human Pro-Inflammatory Panel on an
MSD Imager (Meso Scale Discovery, Rockville, MD,
USA) with minimum sensitivities of 0.396 mg/mL,
0.130 pg/mL, and 0.120 pg/mL, respectively. Finally, a
TBARS assay was completed to measure oxidative stress
using an R&D Parameter (R & D Systems, Minneapolis,
MN, USA) chemical assay, with a minimum sensitivity of
1.10 ng/mL and an intra-assay CV of 4.34%.
Questionnaires
Between each functional test (during the five-minute rest
period), questionnaires were given to participants that
assessed knee pain and mood. The Knee Injury and
Osteoarthritis Outcome Score (KOOS) [23] question-
naire was used to assess the participants’ opinions on
their pain and associated problems with regards to their
KOA. The Brief Pain Inventory (BPI) allowed patients to
report the severity of their pain and the degree to which
their pain interfered with physical function, as well as
providing a location for reporting of pain medications.
BPI pain was computed as a composite score that incor-
porated all reported time points (hours, days, weeks,
months) into consideration, as per scoring directions
[24]. Finally, the Quick Inventory of Depressive
Symptomology was used for the assessment of the
patients’ depressive symptoms [25]. All questionnaires
were scored according to the questionnaire protocols.
Statistical Analysis
Statistical analyses were carried out using IBM SPSS
Statistics, version 24 (IBM Corp, Chicago IL, USA). All
data were first tested for normality. Chi-square tests were
used to examine group differences in sex, and the
Kruskal-Wallis test was used to compare age between
groups. All other variables of interest were found to be
Diet Interventions for Knee Pain 5

Table 4. Statistical results of all analyses

Measure Baseline 12 wk Change Within Group* Time  Diet (Unadjusted)† Post Hoc Analysis‡
Weight, kg 0.001
CTRL 77.64 6 7.07 75.87 6 7.55 0.144
LFD 88.02 6 18.07 81.30 6 16.70 0.001 0.553
LCD 98.53 6 18.56 89.59 6 17.86 0.001 0.072
BMI, kg/m2 0.003

Downloaded from https://academic.oup.com/painmedicine/advance-article-abstract/doi/10.1093/pm/pnz022/5380130 by guest on 14 March 2019

CTRL 26.9 6 3.02 26.6 6 3.46 0.332
LFD 29.65 6 4.48 27.33 6 3.68 0.002 0.788
LCD 35.64 6 7.35 32.53 6 7.11 0.001 0.029
BPI pain 0.283
CTRL 3.64 6 2.35 3.86 6 2.19 0.682
LFD 2.87 6 1.93 2.50 6 1.84 0.267 0.802
LCD 3.06 6 1.73 2.43 6 1.77 0.137 0.474
BPI interference 0.713
CTRL 3.41 6 0.708 2.65 6 1.19 0.113
LFD 2.87 6 1.92 1.65 6 1.84 0.267 0.407
LCD 2.46 6 1.64 0.986 6 1.11 0.022 0.081
KOOS pain 0.893
CTRL 3.23 6 0.542 2.78 6 0.759 0.071
LFD 2.56 6 0.734 2.24 6 0.729 0.337 0.140
LCD 2.68 6 0.360 2.20 6 0.730 0.058 0.150
KOOS QOL 0.605
CTRL 3.67 6 0.718 3.26 6 0.759 0.047
LFD 2.91 60.785 2.08 6 0.954 0.534 0.167
LCD 3.28 6 0.604 2.75 60.627 0.031 0.344
QIDS severity 0.910
CTRL 1.86 61.07 1.43 60.787 0.200
LFD 1.67 6 0.817 1.33 6 0.516 0.175 0.895
LCD 1.63 6 0.744 1.13 6 0.354 0.104 0.652
TS intensity 0.084
CTRL 14.57 6 20.80 25.71 6 21.49 0.344
LFD 26.83 6 36.69 33.33 6 32.66 0.263 0.606
LCD 11.25 6 11.25 –0.6250 6 10.16 0.043 0.307
TS unpleas 0.819
CTRL 22.42 6 29.57 21.42 6 18.64 0.933
LFD 35.5 6 41.9 26.67 6 32.65 0.358 0.706
LCD 14.00 6 13.02 7.500 6 15.11 0.295 0.560
CS intensity 0.454
CTRL 31.42 6 17.72 25.71 6 15.11 0.476
LFD 26.7 6 26.6 16.7 6 18.6 0.447 0.562
LCD 27.50 6 14.14 7.500 6 11.65 0.013 0.218
CS unpleas 0.899
CTRL 34.3 6 23.7 25.71 6 15.11 0.407
LFD 26.7 6 26.6 16.7 6 18.6 0.447 0.491
LCD 25.63 6 15.68 11.25 6 14.57 0.100 0.235
TW intensity 0.066
CTRL 5.71 6 9.75 18.57 6 21.15 0.163
LFD 3.67 6 8.04 –6.67 6 16.32 0.186 0.047
LCD 11.25 6 14.57 3.750 6 7.440 0.265 0.586
TW unpleas 0.283
CTRL 7.86 6 11.49 10.0 6 11.54 0.589
LFD 3.67 6 8.04 –6.67 6 16.32 0.186 0.689
LCD 7.500 6 10.35 3.750 6 7.440 0.476 0.080
IL-6 0.141
CTRL 0.793 6 0.415 1.74 6 2.03 0.329
LFD 0.975 6 0.460 0.965 6 0.377 0.894 0.800
LCD 1.131 6 1.129 1.090 6 0.7045 0.825 0.932
CRP 0.383
CTRL 5.92 6 6.28 3.93 6 3.76 0.578
LFD 2.08 6 2.12 3.91 6 5.38 0.242 0.632
LCD 3.10 6 5.54 2.33 6 2.18 0.605 0.525
TNF-a 0.117
CTRL 2.19 6 0.418 2.58 6 0.572 0.386
LFD 2.42 6 0.497 2.24 6 0.383 0.052 0.989

(continued)
6 Strath et al.

Table 4. continued
Measure Baseline 12 wk Change Within Group* Time  Diet (Unadjusted)† Post Hoc Analysis‡
LCD 3.41 6 0.949 3.35 6 0.839 0.578 0.048
IFN-c 0.277
CTRL 11.3 6 14.7 14.9 6 20.2 0.289
LFD 4.90 6 2.28 4.71 6 1.35 0.854 0.211
LCD 7.44 6 2.90 8.98 6 3.40 0.184 0.500
Leptin 0.688

Downloaded from https://academic.oup.com/painmedicine/advance-article-abstract/doi/10.1093/pm/pnz022/5380130 by guest on 14 March 2019

CTRL 34.67 6 21.53 23.85 6 10.27 0.417
LFD 34.55 6 32.81 24.2 6 20.4 0.110 1.000
LCD 54.9 6 29.6 37.3 6 32.1 0.021 0.462
TBARS 0.002
CTRL 0.656 6 0.149 0.643 6 0.116 0.537
LFD 0.552 6 0.105 0.595 6 0.115 0.397 0.591
LCD 0.691 6 0.205 0.525 6 0.122 0.003 0.816

BMI ¼ body mass index; BPI ¼ Brief Pain Inventory; CRP ¼ C-reactive protein; CS ¼ chair stands; CTRL ¼ control; IFN ¼ interferon; IL ¼ interleukin;
KOOS QOL ¼ Knee Injury and Osteoarthritis Outcome Score quality of life; LCD ¼ low-carbohydrate diet; LFD ¼ low-fat diet; QIDS ¼ Quick Inventory of
Depressive Symptomology; TBARS ¼ thiobarbituric acid reactive substances; TNF ¼ tumor necrosis factor; TS ¼ temporal summation; TW ¼ timed walk.
*T test.
†
Repeated measures analysis of variance.
‡
Dunnett (two-sided).

110 2
Mean (SEM) Weight (kg)

Mean (SEM) Change in

100
1
90
Score
80 0
70
LCD
60 -1
40 LFD
20 CTRL
0 -2 *
BL 3 6 9 12 LCD LFD CTRL
Experimental Week Diet Group

Figure 1. Mean (standard error of the mean) weight (kg) of all Figure 2. Mean (standard error of the mean) change in pain in-
groups (low-carbohydrate diet, top; low-fat diet, middle; con- terference scores on the Brief Pain Inventory between week 12
trol, bottom) over the 12-week intervention period. and baseline for all groups. *P < 0.05.

normally distributed, so a univariate analysis of variance
(ANOVA) was carried out comparing all groups at base-
line to confirm that the groups did not differ at the out-
set. Repeated-measures ANOVAs were used to detect
change over time (BL-12 weeks), and two-sided Dunnett
tests were used to compare the diet groups with the
CTRL. Changes within each group from baseline to 12
weeks were examined by paired t tests. To determine the
impact of change in TBARS on other variables of interest,
we conducted a multivariate ANOVA with TBARS as a
covariate. Leptin is an adipokine released by adipose tis-
sue. Thus, we assessed the change in leptin levels between
the diet groups as a function of weight. A significance
level of P < 0.05 was used in all cases.
Results
Data are reported for the 21 men and women who com-
pleted the study (Table 3). Two participants in the CTRL
group failed to complete the study, and one in the LFD
was lost to nonadherence. As displayed, sex was rela-
tively evenly distributed between the three groups, but
race was weighted more toward non-Hispanic whites
(NHWs). No group differences were found at the base-
line time point for all variables. Paired t tests (Table 4)
revealed a significant decrease in weight (Figure 1) and
BMI in the LCD and LFD groups, but not the CTRL
group. However, the tests also revealed that only in the
LCD group were there significant improvements in pain
interference scores (Figure 2), quality of life (Figure 3),
pain intensity (temporal summation and chair stands)
(Figure 4A, B; Supplementary Data), TBARS (Figure 5A;
Supplementary Data), and leptin (Figure 5B). These dif-
ferences were not evident in the LFD group or the CTRL
group. In addition to the results of the paired t tests,
Table 4 also shows the results of a repeated-measures
ANOVA, confirming a significant time  diet interaction
in the LCD group for TBARS (P ¼ 0.018), but no other
Diet Interventions for Knee Pain 7

interactions. The change in TBARS level correlated with change in weight was related to the change in leptin levels
changes in pain intensity following temporal summation between the groups (F(1, 1 4) ¼ 5.556, P < 0.05).
(r2 ¼ 0.2264, P < 0.05) (Figure 6A), and there was a Ordinal scores and performance measures for TW and
trend suggesting a relation between change in TBARS CS tests are shown in the Supplementary Data for all
and pain intensity associated with the chair stand test groups at baseline and at the end of the 12 weeks.
(r2 ¼ 0.1633, P ¼ 0.090 (Figure 6B). As expected, the

Downloaded from https://academic.oup.com/painmedicine/advance-article-abstract/doi/10.1093/pm/pnz022/5380130 by guest on 14 March 2019

Discussion
0.2
KOA is a widespread public health issue that has many
Mean (SEM) Change in

effects on patients’ physical, cognitive, and emotional

0
functioning. The prevalence rate of KOA is higher in the
older population, wherein pharmacological intervention
Score

-0.2 is often inadequate and carries a greater risk of side

effects. In addition, opioid abuse rates have continued to
-0.4 increase [26], demonstrating the critical need for alterna-
tive forms of pain management. Thus, a holistic, multi-
* modal approach may be a viable alternative for this
LCD LFD CTRL population of chronic pain patients. Based on the as-
Diet Group sumption that oxidative stress may be one of the causes
of the pain experience in patients [27,28], we investigated
Figure 3. Mean (standard error of the mean) change in quality the potential of a dietary intervention to reduce pain in
of life measure in the Knee Injury and Osteoarthritis Outcome
older adults with KOA. Our pilot data showed encourag-
Score for all groups over the 12-week intervention. *P < 0.05.
ing results, whereby our LCD reduced functional pain,

A B
20 30
Mean (SEM) Change in
Mean (SEM) Change in

20
10
10
Score
Score

0 0

-10
-10
-20

-20 -30
*
LCD LFD CTRL LCD LFD CTRL
Diet Group Diet Group

Figure 4. Mean (standard error of the mean) change in intensity ratings for all groups between week 12 and baseline for functional
tests. A) Temporal summation. B) Chair stands. *P < 0.05.

A B
5
Mean (SEM) Change (ng/ml)

0.1
Mean (SEM) Change (mM)

0
0.0
-5

-10
-0.1
-15

-0.2 * -20
LCD LFD CTRL LCD LFD CTRL
Diet Group Diet Group

Figure 5. Mean (standard error of the mean) change in serum concentrations of (A) thiobarbituric acid reactive substances (mM) and
(B) leptin (ng/mL) for all groups between week 12 and baseline. *P < 0.05.
8 Strath et al.
A 0.4
LCD
TBARS (mM)
LFD
CTRL
0.2
-40 -20 20 40 60 80
TS Intensity
-0.2
-0.4
Figure 6. Change in thiobarbituric acid reactive substances (week 12 – baseline, y-axis) as a function of the change in pain intensity
rating (x-axis) following the temporal summation task (A) or the repeated chair stands task (B). Participants are illustrated by group
(CTRL, LFD, LCD). Regression line was calculated using data from all participants. Values below 0 represent a reduction in values at
the end of the intervention when compared to baseline.
leptin, oxidative stress, and self-reported pain measures
while increasing quality of life.
Based on the role that carbohydrates play in oxidative
stress, we hypothesized that excess carbohydrates in the
diet would result in oxidative stress, pain, and inflamma-
tion. Consequently, these effects would decrease if con-
sumption of carbohydrates was also reduced in patients
with KOA. Because there were no significant differences
in weight loss between the diet groups, we believe that
the differences in pain scores between the groups are
unrelated to weight loss, but are an effect of the quality
of the diet. This belief has been reinforced in the litera-
ture in studies of arthritis [29]. Our data show that the
LCD group had dramatically reduced self-reported pain
scores (intensity and unpleasantness) in the chair stands,
temporal summation, and timed walk. Quality of life
scores (KOOS) significantly increased. TBARS decreased
in the blood in the LCD group when compared with the
LFD and CTRL groups. There were no significant differ-
ences in pro-inflammatory cytokine levels over the
12 weeks. This suggests that KOA pain may not be di-
rectly related to peripheral inflammation. However, our
data indicate that oxidative stress may be linked to func-
tional KOA pain as the change in TBARS was related to
the change in pain intensity ratings following temporal
summation and, possibly, repeated chair stands. The re-
duction of TBARS in the LCD group was primarily
driven by women, so further investigation into the sex-
specific effects of diet on TBARS may be warranted.
Nutritional stress caused by poor diet promotes a state
of oxidative stress by increasing systemic free radical
damage and reducing antioxidant status [30]. Recently,
the interaction of AGEs and oxidative stress and the
effects it has on the body have become an extensive field
of interest. Glycation is a nonenzymatic reaction between
excess carbohydrates and other macronutrients such as
B 0.4
TBARS (mM)
0.2
Downloaded from https://academic.oup.com/painmedicine/advance-article-abstract/doi/10.1093/pm/pnz022/5380130 by guest on 14 March 2019
-60 -40 -20 20 40
CS Intensity
-0.2
-0.4
proteins, lipids, and nucleic acids [31,32] that produce
AGEs. AGEs can be ingested and created endogenously
in the presence of excess carbohydrates, as well as
through other cellular processes [33]. Under normal con-
ditions, cells have innate mechanisms that help with de-
toxification and prevent accumulation of AGEs [34].
However, when the system becomes overloaded, these
compounds create a state that ultimately leads to cellular
breakdown [35,36] and cell death. AGEs not only exert
their deleterious actions this way, but also through their
interactions with specific receptors.
Receptor for AGEs (RAGE) is a member of the immu-
noglobulin cell surface receptor family [31,37]. The bind-
ing of AGEs to RAGE stimulates various intracellular
signaling pathways, such as mitogen-activated protein
kinases, extracellular signal–regulated kinases 1 and 2,
cyclin-dependent kinase inhibitor 1, and several other
kinases [38], which leads to apoptosis, inflammation,
and cell differentiation. In addition, stimulation of
RAGE also activates nuclear factor kappa-light-chain-
enhancer of activated B cells (NFkB) and the subsequent
transcription of many pro-inflammatory genes [39,40].
Even more interestingly, RAGE-induced activation of
NFkB is self-perpetuated and maintained through posi-
tive feedback, creating a cycle of self-renewing inflamma-
tory signals and cytokines [39]. RAGE can also directly
induce oxidative stress in several other ways, including
activation of nicotinamide adenine dinucleotide
phosphate-oxidase, decreasing the activity of superoxide
dismutase and catalase pathways [41]. RAGE activation
also reduces cellular antioxidant defenses, such as gluta-
thione and ascorbic acid, which in turn supports the pro-
duction of AGEs [42]. AGEs also directly stimulate
reactive oxygen species (ROS) production [43]. ROS are
pro-inflammatory mediators for immune cells and can
cause cell damage and apoptosis [33]. Thus, it is likely
Diet Interventions for Knee Pain
that poor-quality diet can promote the production of
ROS, activation of immune cells, and, subsequently, in-
flammation that contributes to the development of KOA
and pain.
Although our present study showed promise for the
treatment of KOA with an LCD, there were some limita-
tions. First, given that this was a pilot study, the sample
size was small and not racially diverse, and the distribu-
tion of racial groups was unequal between the diet
groups. Therefore, a larger, more diverse sample would
allow investigations of race and sex, both known to be
differentially affected by pain [44–48] and diet [49–52].
Second, subsequent long-term follow-up assessments will
be able to determine whether participants in the LCD
group maintained their lifestyle change, or a version of it,
when positive results were experienced. Honesty on self-
reporting measures for food intake, long-term adherence,
and acceptance will be an important factor in reducing
the re-emergence of KOA pain as subjects age. Studying
aspects that increase these factors will be critical to wide-
spread adoption of these treatments.
Conclusions
Our exploratory pilot study demonstrates an improve-
ment in self-reported and functional pain in older adults
with KOA following an LCD intervention. In only
12 weeks, the quality of life and functional pain of this
population were significantly improved, which may have
been the result of a reduction in oxidative stress. These
preliminary data are reassuring and support the use of
the LCD as an effective therapeutic approach for older
patients with KOA. Further clinical trials will be per-
formed to ensure a greater sample size and to examine
other factors affecting the outcome such as race/ethnicity
and sex differences. Finally, as oxidative stress is a factor
in many inflammatory conditions, it may be possible to
utilize our LCD intervention for other pain conditions
and disorders.
Acknowledgments
The authors are grateful to Maryellen Williams and
Heather Hunter for serum analyses, Dr. D’Ann Somerall
for assistance with blood draws, and the nursing staff in
the Susan Mott Webb Building at the University of
Alabama at Birmingham.
Supplementary Data
Supplementary data are available at Pain Medicine
online.
References
1. Zhang Y, Jordan JM. Epidemiology of osteoarthritis.
Clin Geriatr Med 2010;26(3):355–69.
9
2. Neogi T. The epidemiology and impact of pain in os-
teoarthritis. Osteoarthritis Cartilage 2013;21
(9):1145–53.
3. Sowers MR, Karvonen-Gutierrez CA. The evolving
role of obesity in knee osteoarthritis. Curr Opin
Rheumatol 2010;22(5):533–7.
4. Cooper C, Snow S, McAlindon TE, et al. Risk factors
Downloaded from https://academic.oup.com/painmedicine/advance-article-abstract/doi/10.1093/pm/pnz022/5380130 by guest on 14 March 2019
for the incidence and progression of radiographic
knee osteoarthritis. Arthritis Rheum 2000;43(5):
995–1000.
5. O’Neil CK, Hanlon JT, Marcum ZA. Adverse effects
of analgesics commonly used by older adults with
osteoarthritis: Focus on non-opioid and opioid
analgesics. Am J Geriatr Pharmacother 2012;10(6):
331–42.
6. Mobasheri A, Henrotin Y. Biomarkers of (osteo)ar-
thritis. Biomarkers 2015;20(8):513–8.
7. Sokolove J, Lepus CM. Role of inflammation in the
pathogenesis of osteoarthritis: Latest findings and
interpretations. Ther Adv Musculoskelet Dis 2013;5
(2):77–94.
8. Herrero-Beaumont G, Roman-Blas JA, Bruyere O,
et al. Clinical settings in knee osteoarthritis:
Pathophysiology guides treatment. Maturitas 2017;
96:54–7.
9. Freedman MR, King J, Kennedy E. Popular diets: A
scientific review. Obes Res 2001;9(Suppl 1):1S–40S.
10. Mittal M, Siddiqui MR, Tran K, Reddy SP, Malik
AB. Reactive oxygen species in inflammation and
tissue injury. Antioxid Redox Signal 2017;20(7):
1126–67.
11. Afari N, Mostoufi S, Noonan C, et al. C-reactive pro-
tein and pain sensitivity: Findings from female twins.
Ann Behav Med 2011;42(2):277–83.
12. Totsch SK, Quinn TL, Strath LJ, et al. The impact of
the Standard American Diet in rats: Effects on behav-
ior, physiology and recovery from inflammatory in-
jury. Scand J Pain 2017;17(1):316–24.
13. Totsch SK, Meir RY, Quinn TL, et al. Effects of a
Standard American Diet and an anti-inflammatory
diet in male and female mice. Eur J Pain 2018; 22
(7):1203–13.
14. Totsch SK, Waite ME, Tomkovich A, et al. Total
Western diet alters mechanical and thermal sensitiv-
ity and prolongs hypersensitivity following complete
freund’s adjuvant in mice. J Pain 2016;17(1):
119–25.
15. Oliviero F, Spinella P, Fiocco U, et al. How the
Mediterranean diet and some of its components mod-
ulate inflammatory pathways in arthritis. Swiss Med
Wkly 2015;145:w14190.
16. Veronese N, Stubbs B, Noale M, et al. Adherence to
the Mediterranean diet is associated with better qual-
ity of life: Data from the Osteoarthritis Initiative. Am
J Clin Nutr 2016;104(5):1403–9.
17. McKellar G, Morrison E, McEntegart A, et al. A pilot
study of a Mediterranean-type diet intervention in
10
female patients with rheumatoid arthritis living in
areas of social deprivation in Glasgow. Ann Rheum
Dis 2007;66(9):1239–43.
18. Sköldstam LB, Hagfors L, Johansson G. Weight re-
duction is not a major reason for improvement in
rheumatoid arthritis from lacto-vegetarian, vegan or
Mediterranean diets. Nutr J 2005;4(15).
19. Guldbrand H, Lindstrom T, Dizdar B, et al.
Randomization to a low-carbohydrate diet advice
improves health related quality of life compared with
a low-fat diet at similar weight-loss in type 2 diabetes
mellitus. Diabetes Res Clin Pract 2014;106(2):221–7.
20. Hagen KB, Byfuglien MG, Falzon L, Olsen SU,
Smedslund G. Dietary interventions for rheumatoid
arthritis. Cochrane Database Syst Rev 2009;(1):
CD006400.
21. Altman R, Asch E, Bloch D, et al. Development of cri-
teria for the classification and reporting of osteoar-
thritis. Classification of osteoarthritis of the knee.
Diagnostic and Therapeutic Criteria Committee of
the American Rheumatism Association. Arthritis
Rheum 1986;29(8):1039–49.
22. Guralnik JM, Simonsick EM, Ferrucci L, et al. A short
physical performance battery assessing lower extrem-
ity function: Association with self-reported disability
and prediction of mortality and nursing home admis-
sion. J Gerontol 1994;49(2):M85–94.
23. Roos EM. Knee Injury and Osteoarthritis Outcome
Score (KOOS). 2003.
24. Cleeland CS, ed. Brief Pain Inventory. Houston, TX:
MD Anderson Cancer Ceter; 1991.
25. Rush AJ, Trivedi MH, Ibrahim HM, et al. The 16-
Item Quick Inventory of Depressive Symptomatology
(QIDS), clinician rating (QIDS-C), and self-report
(QIDS-SR): a psychometric evaluation in patients
with chronic major depression. Biol Psychiatry
2003;54(5):573–83.
26. Palmer RE, Carrell DS, Cronkite D, et al. The preva-
lence of problem opioid use in patients receiving
chronic opioid therapy: Computer-assisted review of
electronic health record clinical notes. Pain 2015;156
(7):1208–14.
27. Battisti E, Albanese A, Guerra L, Argnani L,
Giordano N. Alpha lipoic acid and superoxide dis-
mutase in the treatment of chronic low back pain. Eur
J Phys Rehabil Med 2013;49(5):659–64.
28. Kolberg C, Horst A, Moraes MS, et al. Peripheral oxi-
dative stress blood markers in patients with chronic
back or neck pain treated with high-velocity, low-am-
plitude manipulation. J Manipulative Physiol Ther
2015;38(2):119–29.
29. Skoldstam LBL, Hagfors L, Johansson G. Weight re-
duction is not a major reason for improvement in
rheumatoid arthritis from lacto-vegetarian, vegan or
Mediterranean diets. Nutr J 2005;4:15–20.
30. Newsholme P, Cruzat VF, Keane KN, Carlessi R, de
Bittencourt PI Jr. Molecular mechanisms of ROS
Strath et al.
production and oxidative stress in diabetes. Biochem
J 2016;473(24):4527–50.
31. Gkogkolou P, Bohm M. Advanced glycation end
products: Key players in skin aging?
Dermatoendocrinol 2012;4(3):259–70.
32. Goldin A, Beckman JA, Schmidt AM, Creager MA.
Advanced glycation end products: Sparking the devel-
Downloaded from https://academic.oup.com/painmedicine/advance-article-abstract/doi/10.1093/pm/pnz022/5380130 by guest on 14 March 2019
opment of diabetic vascular injury. Circulation 2006;
114(6):597–605.
33. Bayir H, Kagan VE. Bench-to-bedside review:
Mitochondrial injury, oxidative stress and apoptosis–
there is nothing more practical than a good theory.
Crit Care 2008;12(1):206.
34. Thornalley PJ. The enzymatic defence against glyca-
tion in health, disease and therapeutics: A symposium
to examine the concept. Biochem Soc Trans 2003;31
(6):1341–2.
35. Conner JR, Beisswenger PJ, Szwergold BS. Some clues
as to the regulation, expression, function, and distri-
bution of fructosamine-3-kinase and fructosamine-3-
kinase-related protein. Ann N Y Acad Sci 2005;1043
:824–36.
36. Conner JR, Beisswenger PJ, Szwergold BS. The
expression of the genes for fructosamine-3-kinase
and fructosamine-3-kinase-related protein appears
to be constitutive and unaffected by environmental
signals. Biochem Biophys Res Commun 2004;323(3):
932–6.
37. Chavakis T, Bierhaus A, Nawroth PP. RAGE (recep-
tor for advanced glycation end products): A central
player in the inflammatory response. Microbes Infect
2004;6(13):1219–25.
38. Fleming TH, Humpert PM, Nawroth PP, Bierhaus A.
Reactive metabolites and AGE/RAGE-mediated cel-
lular dysfunction affect the aging process: A mini-re-
view. Gerontology 2011;57(5):435–43.
39. Bierhaus A, Humpert PM, Morcos M, et al.
Understanding RAGE, the receptor for advanced gly-
cation end products. J Mol Med 2005;83
(11):876–86.
40. Miyata T, Taneda S, Kawai R, et al. Identification of
pentosidine as a native structure for advanced glyca-
tion end products in beta-2-microglobulin-containing
amyloid fibrils in patients with dialysis-related amy-
loidosis. Proc Natl Acad Sci U S A 1996;93
(6):2353–8.
41. Furukawa S, Fujita T, Shimabukuro M, et al.
Increased oxidative stress in obesity and its impact on
metabolic syndrome. J Clin Invest 2004;114
(12):1752–61.
42. Ramasamy R, Vannucci SJ, Yan SS, et al. Advanced
glycation end products and RAGE: A common thread
in aging, diabetes, neurodegeneration, and inflamma-
tion. Glycobiology 2005;15(7):16R–28R.
43. Rains JL, Jain SK. Oxidative stress, insulin signal-
ing, and diabetes. Free Radic Biol Med 2011;50
(5):567–75.
Diet Interventions for Knee Pain
44. Berkley KJ. Sex differences in pain. Behav Brain Sci
1997;20(3):371–80; discussion 435–513.
45. Carey TS, Freburger JK, Holmes GM, et al. Race,
care seeking, and utilization for chronic back and
neck pain: Population perspectives. J Pain 2010;11
(4):343–50.
46. Carey TS, Garrett JM. The relation of race to out-
comes and the use of health care services for acute
low back pain. Spine 2003;28(4):390–4.
47. Anderson KO, Green CR, Payne R. Racial and ethnic
disparities in pain: Causes and consequences of un-
equal care. J Pain 2009;10(12):1187–204.
48. Totsch SK, Quinn TL, Strath LJ, et al. The impact of
the Standard American Diet in rats: Effects on behav-
ior, physiology and recovery from inflammatory in-
jury. Scand J Pain 2017;17:316–24.
11
49. Kodama K, Tojjar D, Yamada S, et al. Ethnic differ-
ences in the relationship between insulin sensitivity
and insulin response: A systematic review and meta-
analysis. Diabetes Care 2013;36(6):1789–96.
50. Zhang Q, Wang Y, Huang ES. Changes in racial/eth-
nic disparities in the prevalence of type 2 diabetes by
obesity level among US adults. Ethn Health 2009;14
Downloaded from https://academic.oup.com/painmedicine/advance-article-abstract/doi/10.1093/pm/pnz022/5380130 by guest on 14 March 2019
(5):439–57.
51. Kuk JL, Ross R. Influence of sex on total and regional
fat loss in overweight and obese men and women. Int
J Obes (Lond) 2009;33(6):629–34.
52. Tirosh A, de Souza RJ, Sacks F, et al. Sex differences
in the effects of weight loss diets on bone mineral den-
sity and body composition: POUNDS LOST trial.
J Clin Endocrinol Metab 2015;100(6):2463–71.