Article https://doi.org/10.

1038/s41467-023-41969-1

The Personalized Nutrition Study (POINTS):

evaluation of a genetically informed weight
loss approach, a Randomized Clinical Trial
Received: 26 April 2023 Christoph Höchsmann 1,2 , Shengping Yang2, José M. Ordovás 3,
James L. Dorling 4, Catherine M. Champagne 2, John W. Apolzan 2,
Accepted: 26 September 2023
Frank L. Greenway 2, Michelle I. Cardel5,6, Gary D. Foster 5,7 &
Corby K. Martin 2

Check for updates

1234567890():,;
1234567890():,;

Weight loss (WL) differences between isocaloric high-carbohydrate and high-fat

diets are generally small; however, individual WL varies within diet groups.
Genotype patterns may modify diet effects, with carbohydrate-responsive gen-
otypes losing more weight on high-carbohydrate diets (and vice versa for fat-
responsive genotypes). We investigated whether 12-week WL (kg, primary out-
come) differs between genotype-concordant and genotype-discordant diets. In
this 12-week single-center WL trial, 145 participants with overweight/obesity
were identified a priori as fat-responders or carbohydrate-responders based on
their combined genotypes at ten genetic variants and randomized to a high-fat
(n = 73) or high-carbohydrate diet (n = 72), yielding 4 groups: (1) fat-responders
receiving high-fat diet, (2) fat-responders receiving high-carbohydrate diet, (3)
carbohydrate-responders receiving high-fat diet, (4) carbohydrate-responders
receiving high-carbohydrate diet. Dietitians delivered the WL intervention via 12
weekly diet-specific small group sessions. Outcome assessors were blind to diet
assignment and genotype patterns. We included 122 participants (54.4 [SD:13.2]
years, BMI 34.9 [SD:5.1] kg/m2, 84% women) in the analyses. Twelve-week WL did
not differ between the genotype-concordant (−5.3 kg [SD:1.0]) and genotype-
discordant diets (−4.8 kg [SD:1.1]; adjusted difference: −0.6 kg [95% CI: −2.1,0.9],
p = 0.50). With the current ability to genotype participants as fat- or carbohy-
drate-responders, evidence does not support greater WL on genotype-
concordant diets. ClinicalTrials identifier: NCT04145466.

The 2017–2018 National Health and Nutrition Examination Survey conditions and premature mortality2, presenting public health and
(NHANES) showed that almost 43% of US adults aged 20 and over have economic challenges3,4.
obesity, including 9.0% with severe obesity, and another 31% are Many weight loss (WL) strategies emphasize either high-
overweight1. Excess body fat increases the risk of numerous medical carbohydrate (and low-fat) or high-fat (low-carbohydrate) diets5,6. WL

1
Department of Health and Sport Sciences, TUM School of Medicine and Health, Technical University of Munich, Munich, Germany. 2Pennington Biomedical
Research Center, Baton Rouge, LA, USA. 3Tufts University, Boston, MA, USA. 4Human Nutrition, School of Medicine, Dentistry and Nursing, College of Medical,
Veterinary and Life of Sciences, University of Glasgow, Glasgow, UK. 5WW International, Inc., New York, NY, USA. 6Department of Health Outcomes and
Biomedical Informatics, University of Florida College of Medicine, Gainesville, FL, USA. 7Center for Weight and Eating Disorders, Perelman School of Medicine,
University of Pennsylvania, Philadelphia, PA, USA. e-mail: [email protected]

Nature Communications | (2023)14:6321 1

Article https://doi.org/10.1038/s41467-023-41969-1

differences between isocaloric high-carbohydrate and high-fat diets are high-carbohydrate diets, may better define fat- and carbohydrate-
generally small or negligible7; however, individual WL varies sub- responsive genetic predisposition scores.
stantially within diet groups6, suggesting that individuals react differ- The present RCT tested the hypothesis that participants assigned
ently to high-carbohydrate or high-fat diets. Retrospective data suggest to a diet corresponding to their a priori-determined (fat-responsive or
that participants with carbohydrate-responsive polymorphisms lose carbohydrate-responsive) genotype would lose more weight over
more weight on high-carbohydrate vs. high-fat diets and vice versa for 12 weeks than those assigned to a diet discordant with their genotype.
those with fat-responsive polymorphisms8. However, these results have Further, we aimed to analyze those with a fat-responsive genotype
not been confirmed in randomized controlled trials (RCT), and the (subsequently “fat-responders”) and carbohydrate-responsive geno-
approach of determining low-fat- and low-carbohydrate-responsive type (subsequently “carbohydrate-responders”) separately. We hypo-
genotypes based on single-nucleotide polymorphisms (SNPs) from thesized that (1) fat-responders would lose more weight on the high-fat
three genes (PPARG, ADRB2, and FABP2)8,9 has been criticized10. Overall, vs. high-carbohydrate diet and conversely (2) carbohydrate-
reports show that most genotype × diet interactions are not significant, responders would lose more weight on the high-carbohydrate vs.
and replication is rare11. A more comprehensive and informative risk high-fat diet. A secondary objective of the present RCT was to test the
score (determined a priori), comprised of a greater number of SNPs newly-developed genetic risk score to determine fat- and
with demonstrated and validated effects on the responses to high-fat/ carbohydrate-responsive genotypes that was based on the current

Completed web screen (n=1392)

Participants could bypass web
screen and complete phone screen Ineligible (n=850)
only. Web screen number is Weight instability (n=198)
independent of the number of Unable to contact (n=188)
phone screens completed/reported . Diabetes (n=186)
BMI (n=122)
Unwilling to follow diet (n=39)
Changed mind (n=36)
Smoker (n=36)
Medical/medications (n=25)
Multiple study criteria (n=16)
Pregnant/breastfeeding (n=4)
Completed phone screen (n=690)
Ineligible (n=385)
Inclusion/exclusion (n=112)
COVID-19 suspension (n=94)
Changed mind (n=59)
Never scheduled (n=35)
Medical/medications (n=35)
Unable to contact (n=24)
Time commitment (n=20)
In another study (n=3)
Weight instability (n=3)
Completed orientation visit (n=305)
Excluded (n=153)
Genetic data (n=106)
Medical/medications (n=20)
No show/lost contact (n=13)
BMI (n=10)
Changed mind (n=4)
Completed baseline visit (n=152)
Excluded (n=7)
Elevated glucose (n=5)
Unable to contact (n=2)

Randomized (n=145)

High-fat diet (n=73) High-carbohydrate diet (n=72)

Lost to follow-up (n=3) Lost to follow-up (n=6)

Discontinued intervention (n=2) Discontinued intervention (n=5)
Family emergency (n=1) Family emergency (n=2)
Medical reasons (n=1) Refused intervention (n=2)
Work conflict (n=1)

Completed study (n=68) Completed study (n=61)

Included in primary analysis (n=65) Included in primary analysis (n=57)

Excluded (n=3) Excluded (n=4)
Genotyping error (n=3) a Genotyping error (n=2) a
No EOS lab weight (n=2) b

Fig. 1 | CONSORT diagram illustrating the flow of participants through the This was reported to the IRB, and, as part of the resolution, their data were removed
POINTS trial. aAn error in the algorithm to determine carbohydrate- and fat- from the dataset. bThese participants were unable to attend the W12 visit in person
responsive genotypes led to the incorrect classification of these participants. These and only completed surveys and questionnaires remotely.
participants were erroneously enrolled as they did not meet the eligibility criteria.

Nature Communications | (2023)14:6321 2

Article https://doi.org/10.1038/s41467-023-41969-1

state-of-the-art in nutrigenomics. We also aimed to determine asso- (54.4 [SD: 13.2] years, BMI 34.9 [SD: 5.1] kg/m2, 84% women, 68%
ciations between baseline insulin levels and homeostatic model White) are provided in Table 1. A comparison of baseline character-
assessment for insulin resistance (HOMA-IR) and differential WL istics between non-completers (n = 16) and completers (n = 122) is
between the diets. These analyses were pursued as previous results provided in Supplementary Table 3.
were mixed with some studies finding that insulin resistance12,13 and
glucose-stimulated insulin secretion14 influenced differential weight Change in the primary outcome
loss between low-fat and low-carbohydrate diets. In contrast, others Weight change did not differ between genotype-concordant (−5.3 kg
found no interaction between glucose-stimulated insulin secretion and [SD: 1.0]) and genotype-discordant diets (−4.8 kg [SD: 1.1]; adjusted
diet type on 12-month weight loss9. Finally, we examined the diet difference: −0.6 kg [95% CI: −2.1, 0.9, p = 0.50]; Table 2, Fig. 2). Among
effects on eating attitudes and behaviors to help elucidate the fat-responders, weight change did not differ between the high-fat
mechanisms by which any observed differences in WL occurred. As (−5.5 kg [SD: 1.2]) and the high-carbohydrate diet (−5.3 kg [SD: 1.3];
program adherence diminishes over time15, we chose a 12-week inter- adjusted difference: −0.2 kg [95% CI: −2.1, 1.6, p = 0.78]; Table 2).
vention period, which generally has lower attrition (~19%) than 6- Similarly, among carbohydrate-responders, weight change did not
(~35%) and 12-month (~54%) programs16, and short-term WL is asso- differ between the high-carbohydrate (−5.1 kg [SD: 1.6]) and high-fat
ciated with long-term results17,18. diet (−4.1 kg [SD: 1.7]; adjusted difference: −1.3 kg [95% CI: −3.9, 1.3,
p = 0.49]; Table 2). Raw differences are presented in Supplementary
Results Table 5.
Figure 1 shows the flow of participants through the study. Of the
2082 participants who screened for the study, 305 were eligible Percent weight change and change in body fat and body
following the web/phone screen and were invited to the orientation composition
visit. After eligibility verification based on medical history, medica- Similar to absolute weight change, percent weight change (adjusted
tion inventory, and physical measures, 275 remained and completed difference: −0.6% [95% CI: −2.1, 0.9, p = 0.61]) and change in body
a genealogy test. Of these 275 individuals, 106 (~39%) were excluded fat (adjusted difference: −0.5% [95% CI: −2.4, 1.4]) did not differ
because they had a genotype that was classified as responsive to between genotype-concordant and genotype-discordant diets
neither a high-fat nor a high-carbohydrate diet or as responsive to (Table 2, Fig. 2). Among fat-responders, percent weight change
both diets. Of the remaining 169 individuals, 112 (~41%) were fat- (adjusted difference: −0.2% [95% CI: −2.1, 1.7, p = 0.83]) and change in
responders, and 57 (~20%) were carbohydrate-responders. Before the body fat (adjusted difference: 0.9% [95% CI: −1.3, 3.0]) did not differ
baseline visit (completed by 152 participants), 17 participants were between the high-fat and the high-carbohydrate diet (Table 2). Simi-
excluded because we either lost contact between the orientation visit larly, among carbohydrate-responders, percent weight change
and the baseline visit (n = 13) or because participants changed their (adjusted difference: −1.2% [95% CI: −4.2, 1.7, p = 0.57]) and change in
minds about willingness to participate (n = 4). Following the baseline body fat (adjusted difference: −3.4% [95% CI: −7.5, 0.8]) did not differ
visit, 7 additional participants were excluded due to elevated glucose between the high-carbohydrate and high-fat diet (Table 2). Changes in
levels (n = 5) or lost contact (n = 2). Of the 145 participants rando- waist circumference (adjusted difference: −0.5 cm [95% CI: −2.3, 1.3]),
mized, 16 were lost to follow-up (W12), and 129 completed the trial. hip circumference (adjusted difference: −1.0 cm [95% CI: −3.6, 1.6]),
Seven participants were excluded from the analyses because they and waist-hip ratio (adjusted difference: 0.00 [95% CI: −0.02, 0.03]) did
were incorrectly genotyped and erroneously enrolled (n = 5; removal not differ between genotype-concordant and genotype-discordant
from dataset suggested by IRB) or failed to provide weight data at diets (Table 2). Raw differences are presented in Supplementary
W12 (n = 2). Baseline characteristics of all 122 included participants Table 5.

Table 1 | Participant characteristics

Fat-responders (n = 85) Carbohydrate-responders (n = 37)

All participants (N = 122) High-fat diet (n = 44) High-carbohydrate diet (n = 41) High-fat diet (n = 21) High-carbohydrate diet (n = 16)
Race, n (%)
White 83 (68.0) 30 (68.2) 31 (75.6) 12 (57.1) 10 (62.5)
Black/ African American 36 (29.5) 12 (27.3) 10 (24.4) 9 (42.9) 5 (31.2)
Other 3 (2.5) 2 (4.5) 0 (0.0) 0 (0.0) 1 (6.2)
Sex, n (%)
Female 102 (83.6) 37 (84.1) 35 (85.4) 17 (81.0) 13 (81.2)
Male 20 (16.4) 7 (15.9) 6 (14.6) 4 (19.0) 3 (18.8)
Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD)
Age, years 54.4 (13.2) 57.4 (11.5) 54.4 (14.2) 49.8 (14.1) 52.4 (13.0)
Weight, kg 94.3 (15.2) 94.2 (14.0) 93.5 (14.4) 95.2 (17.6) 95.6 (18.1)
BMI, kg/m2 34.9 (5.1) 35.1 (5.0) 34.3 (4.8) 35.8 (5.8) 34.8 (5.3)
Body fat, % 45.1 (9.3) 45.0 (9.4) 45.2 (8.5) 43.8 (11.6) 46.1 (8.4)
Waist circumference, cm 109.0 (12.2) 109.3 (11.8) 108.5 (12.3) 109.3 (11.8) 109.2 (14.8)
Hip circumference, cm 118.9 (12.2) 117.5 (10.7) 118.3 (12.1) 120.1 (12.5) 122.8 (15.8)
Waist-hip ratio 0.92 (0.08) 0.94 (0.09) 0.92 (0.08) 0.91 (0.06) 0.89 (0.10)
SBP, mmHg 121.7 (11.9) 120.5 (11.6) 124.1 (12.9) 121.7 (11.5) 119.4 (10.5)
DBP, mmHg 74.7 (7.4) 75.1 (7.0) 74.5 (7.9) 75.0 (6.1) 73.9 (9.3)
Data are mean (SD) for continuous and n (%) for categorical variables.
BMI body mass index, DBP diastolic blood pressure, SBP systolic blood pressure, SD standard deviation.

Nature Communications | (2023)14:6321 3

Article https://doi.org/10.1038/s41467-023-41969-1

Table 2 | Change in weight (kg and %), percent body fat, body composition, and blood pressure during the 12-week inter-
vention in those assigned to a diet concordant vs. discordant with the genotype

All participants Genotype-concordant diet (n = 60) Genotype-discordant diet (n = 62)

Mean (SD) Mean (SD) Adjusted differencea (95% CI) p-value
Weight change, kg −5.3 (1.0) −4.8 (1.1) −0.6 (−2.1, 0.9) 0.501
Weight change, % −5.8 (1.0) −5.4 (1.1) −0.6 (−2.1, 1.0) 0.605
Change in body fat, %b −1.3 (1.2) −0.8 (1.3) −0.5 (−2.4, 1.4)
Waist circumference, cm −4.8 (1.1) −4.3 (1.2) −0.5 (−2.3, 1.3)
Hip circumference, cm −4.6 (1.7) −3.7 (1.8) −1.0 (−3.6, 1.6)
Waist-hip ratio 0.01 (0.00) 0.01 (0.00) 0.00 (−0.02, 0.03)
SBP, mmHg 1.2 (2.7) −2.9 (2.9) 4.7 (0.5, 8.8)
DBP, mmHg 0.4 (1.7) 1.0 (1.9) −0.1 (−2.8, 2.5)
Fat-responders High-fat diet (n = 44) High-carbohydrate diet (n = 41)
Mean (SD) Mean (SD) Adjusted differencea (95% CI) p-value
Weight change, kg −5.5 (1.2) −5.3 (1.3) −0.2 (−2.1, 1.6) 0.779
Weight change, % −5.9 (1.3) −5.7 (1.4) −0.2 (−2.1, 1.7) 0.831
Change in body fat, %c −1.1 (1.4) −1.9 (1.6) 0.9 (−1.3, 3.0)
Waist circumference, cm −5.0 (1.4) −4.4 (1.5) −0.6 (−2.7, 1.5)
Hip circumference, cm −3.9 (1.5) −4.0 (1.7) 0.2 (−2.1, 2.6)
Waist-hip ratio 0.00 (0.00) 0.00 (0.00) −0.01 (−0.03, 0.02)
SBP, mmHg 4.5 (3.2) −1.2 (3.5) 6.9 (2.0, 11.8)
DBP, mmHg 1.7 (2.2) 2.9 (2.4) −0.5 (−3.8, 2.9)
Carbohydrate- High-carbohydrate diet (n = 16) High-fat diet (n = 21)
responders
Mean (SD) Mean (SD) Adjusted differencea (95% CI) p-value
Weight change, kg −5.1 (1.6) −4.1 (1.7) −1.3 (−3.9, 1.3) 0.487
Weight change, % −5.7 (1.8) −4.8 (1.9) −1.2 (−4.2, 1.7) 0.565
Change in body fat, % d −1.9 (2.5) 1.4 (2.7) −3.4 (−7.5, 0.8)
Waist circumference, cm −4.4 (2.1) −4.2 (2.3) −0.3 (−3.9, 3.3)
Hip circumference, cm −6.4 (4.3) −2.7 (4.7) −3.9 (−11.1, 3.3)
Waist-hip ratio 0.00 (0.00) 0.00 (0.00) 0.03 (−0.02, 0.08)
SBP, mmHg −5.8 (4.6) −7.2 (5.0) 0.3 (−7.4, 8.0)
DBP, mmHg −2.0 (2.8) −3.0 (3.0) 0.9 (−3.7, 5.5)
CI confidence interval, DBP diastolic blood pressure, SBP systolic blood pressure, SD standard deviation.
a
Mixed-effect model, adjusted for sex, race, and baseline value of the outcome for all data.
b
Data available for 58 of 60 participants (genotype-concordant diet) and 60 of 62 participants (genotype-discordant diet).
c
Data available for 42 of 44 participants (high-fat diet) and 40 of 41 participants (high-carbohydrate diet).
d
Data available for 16 of 16 participants (high-carbohydrate diet) and 20 of 21 participants (high-fat diet).

Change in blood pressure Change in food cravings, appetitive traits, and food preferences
Changes in resting systolic blood pressure (SBP) and DBP did not differ Changes in food cravings did not differ between the genotype-
between genotype-concordant and genotype-discordant diets (SBP concordant and genotype-discordant diets (Table 3). Among
adjusted difference: 4.7 mmHg [95% CI: 0.5, 8.8]; DBP adjusted differ- carbohydrate-responders, those on a high-fat diet decreased
ence: −0.1 mmHg [95% CI: −2.8, 2.5]; Table 2, Fig. 3). Similarly, changes cravings for carbohydrates/starches relative to those on the high-
in SBP and DBP did not differ between the high-fat and the high- carbohydrate diet with an adjusted difference of −0.7 (95% CI: −1.1, −0.4,
carbohydrate diet among fat-responders (SBP difference: 6.9 mmHg p = 0.006, without Holm-Bonferroni adjustment p = 0.001). Changes in
[95% CI: 2.0, 11.8]; DBP difference: −0.5 mmHg [95% CI: −3.8, 2.9]) or all other food cravings did not differ between diets among
between the high-carbohydrate and high-fat diet among carbohydrate carbohydrate-responders (Table 3). Among fat-responders, changes in
responders (SBP difference: 0.3 mmHg [95% CI: −7.4, 8.0]; DBP dif- food cravings did not differ between diets (Table 3). Raw differences are
ference: 0.9 mmHg [95% CI: −3.7, 5.5]; Table 2). Raw differences are presented in Supplementary Table 6. Changes in restraint, disinhibition,
presented in Supplementary Table 5. and hunger (via EI), and food preferences (FPQ) did not differ between
genotype-concordant and genotype-discordant diets (Table 4). Raw
Association between insulin levels and HOMA-IR and weight loss differences are presented in Supplementary Table 7 and baseline scores
Baseline insulin levels (β = −0.036 [95% CI: −0.125, 0.053, p = 0.43]) and in these instruments are reported in Supplementary Table 4.
HOMA-IR (β = −0.165 [95% CI: −0.505, 0.175, p = 0.34]) were not asso-
ciated with weight change (Supplementary Figure 1). There was no diet Diet personalization and intervention satisfaction
× baseline HOMA-IR interaction on weight change (p = 0.37). Similarly, Diet preference (via Diet Personalization Survey, Table 5) and inter-
there was no significant diet × baseline HOMA-IR interaction among vention satisfaction (Table 6) did not differ between the genotype-
carbohydrate-responders (p = 0.62) or fat-responders (p = 0.23; Sup- concordant and genotype-discordant diets. Raw differences are pre-
plementary Fig. 2). sented in Supplementary Table 8.

Nature Communications | (2023)14:6321 4

Article https://doi.org/10.1038/s41467-023-41969-1

Fig. 2 | Change in weight and percent body fat during the 12-week intervention. (n = 21); d genotype-concordant group (n = 58), genotype-discordant group
Results are presented as boxplots for all participants (a, d), as well as for fat- (n = 60); e high-fat diet (n = 42), high-carbohydrate diet (n = 40); f high-carbohy-
responders (b, e) and carbohydrate responders (c, f) separately. a Genotype-con- drate diet (n = 16), high-fat diet (n = 20). In the boxplots, the center line denotes the
cordant group (n = 60, genotype-discordant group (n = 62); b high-fat diet (n = 44), median value (50th percentile), the bounds of the box represent the 25th and 75th
high-carbohydrate diet (n = 41); c high-carbohydrate diet (n = 16), high-fat diet percentiles of the dataset, and the whiskers mark the 5th and 95th percentiles.

Diet adherence among fat-responders on a high-fat diet) that required hospitalization

Adherence to the assigned diets is shown in Fig. 4. We encountered (unrelated to study).
difficulties in obtaining the adherence data from participants due, in
part, to the pandemic and needing to move to remote intervention Discussion
delivery. Consequently, these adherence data are only available for 22 The present RCT determined the participant’s (fat-responsive or car-
of 57 participants (39%) on the high-carbohydrate diet and for 43 of 65 bohydrate-responsive) genotype a priori via a comprehensive genetic
participants (66%) on the high-fat diet (the discrepancy in the percent risk score based on published and validated effects and tested the
complete/missing is noted, though we have no reason to believe that it effects of a genotype-concordant diet on WL over 12 weeks. We found
was systematic). On average, participants on the high-carbohydrate no difference in WL between individuals on the genotype-concordant
diet reported consuming 63.4% (SD: 2.3) of their energy from carbo- vs. genotype-discordant diet. Further, insulin levels or HOMA-IR were
hydrates (target 65%), 20.9% (SD: 2.4) from fat (target 20%), and 16.0% not associated with WL. Food cravings tended to decrease among
(SD: 1.0) from protein (target 15%) in week 4, 63.3% (SD: 2.8) from carbohydrate-responders on a high-fat diet compared to those on a
carbohydrates, 20.5% (SD: 1.7) from fat, and 15.9% (SD: 1.0) from pro- high-carbohydrate diet. Finally, fat-responders on a high-carbohydrate
tein in week 8, and 62.7% (SD: 4.0) from carbohydrates, 20.5% (SD: 2.5) diet tended to decrease resting SBP.
from fat, and 15.7% (SD: 1.8) from protein in week 12. Participants on The lack of significant and clinically meaningful differences in WL
the high-fat diet reported consuming on average 45.4% (SD: 2.2) of (~0.6 kg) between genotype-concordant and genotype-discordant
their energy from carbohydrates (target 45%), 39.4% (SD: 2.0) from fat diets aligns with the literature9,11. In contrast to the well-conducted
(target 40%), and 15.8% (SD: 1.2) from protein (target 15%) in week 4, Gardner et al. study (non-significant difference in WL of 0.7 kg over
44.7% (SD: 2.2) from carbohydrates, 40.5% (SD: 2.1) from fat, and 15.7% 12 months)9, who defined fat vs. carbohydrate-responsive genotypes
(SD: 2.3) from protein in week 8, and 44.5% (SD: 3.4) from carbohy- based on 3 SNPs that were predictive in a preliminary retrospective
drates, 39.9% (SD: 2.5) from fat, and 16.1% (SD: 3.3) from protein in analysis8, we determined fat- or carbohydrate-responsive genotypes
week 12. based on an algorithm involving 10 SNPs. Supported by a recent-meta-
analysis (8 trials with 91 SNPs and 63 genetic loci)11, our results suggest
Session attendance and adverse events that with the current ability to genotype individuals as fat or carbo-
Weekly attendance was similar across the four genotype-diet groups hydrate-responders, there is no evidence that genotype-concordant
(Supplementary Table 9), with weekly session attendance ranging diets result in greater WL.
from 85% to 100%. There were 4 adverse or serious adverse events in Our sample consisted of substantially fewer carbohydrate-
total. Two adverse events occurred among fat-responders on a high- responders (n = 37) than fat-responders (n = 85). We did not limit
carbohydrate diet (unrelated to the study), and there were 2 serious recruitment to achieve equal numbers of participants in each genotype-
adverse events (1 among fat-responders on a high-carbohydrate diet, 1 diet group, and this distribution reflects the prevalence in our

Nature Communications | (2023)14:6321 5

Article https://doi.org/10.1038/s41467-023-41969-1

Fig. 3 | Change in systolic and diastolic blood pressure during the 12-week carbohydrate responders (c, f; high-carbohydrate diet, n = 16, high-fat diet, n = 21)
intervention. Results are presented as boxplots for all participants (a, d; genotype- separately. In the boxplots, the center line denotes the median value (50th
concordant group, n = 60, genotype-discordant group, n = 62), as well as for fat- percentile), the bounds of the box represent the 25th and 75th percentiles of the
responders (b, e; high-fat diet, n = 44, high-carbohydrate diet, n = 41) and dataset, and the whiskers mark the 5th and 95th percentiles.

population. As reported in the Results section, 275 individuals com- WL can reduce food cravings, particularly for foods restricted on
pleted a genealogy test, of which ~39% had a genotype classified as specific diets22, contributing to the hypothesis that food cravings are a
responsive to both or neither of the two diets, ~41% were fat-responders conditioned expression of hunger due to stimuli paired with eating
and ~20% were carbohydrate-responders. Notably, these numbers are certain foods23. Consequently, cravings can be reduced by eliminating
somewhat different from what we had estimated during the study’s or restricting the intake of craved foods. This hypothesis is partially
planning phase, as we expected 1/3 of people to be fat-responders, 1/3 supported by our results as, among carbohydrate-responders, crav-
carbohydrate-responders, and 1/3 to respond to neither or both of the ings tended to decrease for high-carbohydrate foods on the high-fat
specified diets. Future studies with larger samples should verify if this diet. Nonetheless, cravings also decreased modestly for high-fat foods,
uneven distribution between carbohydrate-responders and fat- which is to be expected as the amount of all foods was restricted, and
responders is representative of the general population and further cravings for specific foods correlate with each other24.
investigate the potential effect on WL among carbohydrate-responders. Among fat-responders, a high-carbohydrate diet tended to
Future studies could also consider assigning participants to decrease resting SBP. Nonetheless, these individuals had the highest
genotype-concordant diets without specific energy intake targets and mean SBP of the 4 genotype-diet groups at baseline. Thus, this effect
examine the diet effects not only on WL but also on cardiovascular risk could be explained, in whole or partially, by regression to the mean.
factors. Previously, a low-carbohydrate diet without energy intake Also, all 4 genotype-diet groups had relatively well-controlled
target resulted in greater improvements in body composition, blood blood pressure, leaving little room for improvement through dietary
lipids, and estimated 10-year coronary heart disease risk compared to a changes, making the non-significant improvements potentially more
low-fat diet19. It would be insightful to investigate whether genotype meaningful.
plays a role in cardiovascular risk reduction following a low- This trial has some limitations. First, the genetic algorithm
carbohydrate vs. low-fat diet without calorie restriction. to classify individuals as fat- or carbohydrate-responders was created
Fasting insulin levels and HOMA-IR did not predict WL. Previous based on published literature25–38. However, these (mostly retro-
studies reporting a diet × fasting insulin interaction for WL found lower spective) studies generally had modest sample sizes, and some of
carbohydrate diets to be superior for individuals with greater insulin the genotype × diet interactions, which may be false positives, have
resistance13 and high baseline insulin secretion (30 min after a 75 g oral not been independently replicated. Further, WL is determined by
glucose tolerance test)20, presumably due to a reduced burden on multiple modifiable and non-modifiable (e.g., genetic) factors,
insulin-mediated glucose disposal. However, these studies involved and current knowledge accounts for a small percentage of the
relatively small sample sizes, and findings of the influence of insulin variability. Further genotypes may have influenced participants’ WL
sensitivity21 and insulin secretion9,14 on WL via a low-fat vs. a low- responses in directions different from those predicted from the
carbohydrate diet are inconsistent. measured genotypes. More comprehensive knowledge of the role of

Nature Communications | (2023)14:6321 6

Article https://doi.org/10.1038/s41467-023-41969-1

Table 3 | Changes in food cravings (via the Food Craving Inventory) during the 12-week intervention in those assigned to a diet
concordant vs. discordant with the genotype

All participants Genotype-concordant diet (n = 60) Genotype-discordant diet (n = 62)

Mean (SD) Mean (SD) Adjusted differencea (95% CI)
High fats b
−0.3 (0.1) −0.4 (0.2) 0.1 (−0.1, 0.4)
Sweetsc −0.3 (0.2) −0.5 (0.2) 0.2 (−0.1, 0.4)
Carbohydrates/Starchesd −0.1 (0.2) −0.4 (0.2) 0.3 (0.0, 0.5)
Fast-food fatse −0.3 (0.2) −0.4 (0.2) 0.1 (−0.2, 0.4)
Fruits and vegetablesf −0.1 (0.2) −0.4 (0.2) 0.2 (−0.1, 0.5)
Total cravingsg −0.2 (0.1) −0.4 (0.1) 0.2 (−0.1, 0.4)
Fat-responders High-fat diet (n = 44) High-carbohydrate diet (n = 41)
Mean (SD) Mean (SD) Adjusted differencea (95% CI)
High fats b
−0.4 (0.2) −0.3 (0.2) −0.1 (−0.3, 0.3)
Sweetsc −0.4 (0.2) −0.6 (0.2) 0.2 (−0.1, 0.5)
Carbohydrates/Starchesd −0.2 (0.2) −0.3 (0.2) 0.1 (−0.3, 0.4)
Fast-food fatse −0.4 (0.2) −0.3 (0.3) −0.1 (−0.4, 0.3)
Fruits and vegetablesf −0.3 (0.2) −0.4 (0.3) 0.1 (−0.3, 0.5)
Total cravingsg −0.3 (0.2) −0.3 (0.2) 0.0 (−0.3, 0.3)
Carbohydrate-responders High-carbohydrate diet (n = 16) High-fat diet (n = 21)
Mean (SD) Mean (SD) Adjusted differencea (95% CI)
High fatsb −0.2 (0.2) −0.7 (0.2) 0.5 (0.1, 0.9)
Sweetsc −0.1 (0.2) −0.3 (0.3) 0.2 (−0.2, 0.6)
Carbohydrates/Starchesd 0.1 (0.2) −0.7 (0.2) 0.7 (0.4, 1.1)
Fast-food fatse −0.1 (0.3) −0.5 (0.3) 0.5 (0.0, 1.0)
Fruits and vegetablesf 0.3 (0.3) −0.3 (0.3) 0.6 (0.1, 1.1)
Total cravingsg 0.0 (0.2) −0.5 (0.2) 0.5 (0.2, 0.9)
CI confidence interval, SD standard deviation.
a
Adjusted for sex, race, and baseline value of the outcome.
b
Genotype-concordant diet: 55/60 participants; genotype-discordant diet: 60/62 participants. Fat-responders: 41/44 participants (high-fat diet) and 40/41 participants (high-carbohydrate diet).
Carbohydrate-responders: 14/16 participants (high-carbohydrate diet) and 20/21 participants (high-fat diet).
c
Genotype-concordant diet: 59/60 participants; genotype-discordant diet: 60/62 participants. Fat-responders: 43/44 participants (high-fat diet) and 40/41 participants (high-carbohydrate diet).
Carbohydrate-responders: 16/16 participants (high-carbohydrate diet) and 20/21 participants (high-fat diet).
d
Genotype-concordant diet: 59/60 participants; genotype-discordant diet: 61/62 participants. Fat-responders: 44/44 participants (high-fat diet) and 40/41 participants (high-carbohydrate diet).
Carbohydrate-responders: 15/16 participants (high-carbohydrate diet) and 21/21 participants (high-fat diet).
e
Genotype-concordant diet: 58/60 participants; genotype-discordant diet: 61/62 participants. Fat-responders: 43/44 participants (high-fat diet) and 40/41 participants (high-carbohydrate diet).
Carbohydrate-responders: 15/16 participants (high-carbohydrate diet) and 20/21 participants (high-fat diet).
f
Genotype-concordant diet: 58/60 participants; genotype-discordant diet: 60/62 participants. Fat-responders: 43/44 participants (high-fat diet) and 40/41 participants (high-carbohydrate diet).
Carbohydrate-responders: 15/16 participants (high-carbohydrate diet) and 20/21 participants (high-fat diet).
g
Genotype-concordant diet: 54/60 participants; genotype-discordant diet: 67/62 participants. Fat-responders: 41/44 participants (high-fat diet) and 38/41 participants (high-carbohydrate diet).
Carbohydrate-responders: 13/16 participants (high-carbohydrate diet) and 19/21 participants (high-fat diet).

genetics in WL is needed and should be obtained from genome-wide HOMA-IR has limited sensitivity due to its reliance on fasting insulin
association studies; however, the sample size and experimental and glucose levels and it does not reflect differences between tissues
design required to generate that essential information are beyond (e.g., adipose, muscle) or postprandial physiology. Non-fasting
reach at this time. Additional limitations of the present study include methods yield greater variability of the glucose/insulin dynamics
the relatively small sample size, single-center design, and short time and may have been more suitable. Additionally, the assessment of
frame. A longer timeframe (6–12-month follow-up) may have percent body fat via BIA is a limitation as BIA does not provide
increased the amount and differential weight loss between diets. A information on body fat distribution. Finally, participation in “nutri-
larger sample size might have also allowed for detecting differences genomics” studies generally induces improved diet adherence39–42,
in clinically important secondary outcomes such as changes in independent of the specific recommendations. Therefore, in our
body fat and SBP. Further, we did not provide meals in this study, participants may have responded better to their assigned diets
study, which may have affected dietary adherence (high-fat vs. regardless of their genotype matching, obscuring the specific nutri-
high-carbohydrate). However, this choice was made by design, as our genomics effects.
study was designed as a (pragmatic) effectiveness trial with real- In conclusion, in this 12-week RCT, there was no difference in WL
world conditions rather than an efficacy trial. Additionally, the between individuals with an a priori determined fat- or carbohydrate-
adherence data (albeit limited) suggests that diet adherence was responsive genotype on a high-carbohydrate vs. high-fat diet with
overall satisfactory. In addition to assessing diet adherence con- specific energy targets and the same level of energy restriction
tinuously throughout the study, future studies should also assess the across diets.
macronutrient composition of participants’ habitual diets to see
any differences in the magnitude of the shifts from baseline to the Methods
high-fat or high-carbohydrate diet. Further, when assessing a Design and participants
potential effect modification by insulin resistance status, using an The Personalized Nutrition Study (POINTS, ClinicalTrials.gov identifier:
oral glucose tolerance test (AUC or INS-30) rather than HOMA-IR to NCT04145466) was a 12-week, single-site, parallel-arm WL trial that was
quantify insulin resistance might have been a better option, as approved by the institutional review board (IRB FWA 00006218) of the

Nature Communications | (2023)14:6321 7

Article https://doi.org/10.1038/s41467-023-41969-1

Table 4 | Change in restraint, disinhibition, and hunger and in food preferences during the 12-week intervention in those
assigned to a diet concordant vs. discordant with the genotype

All participants Genotype-concordant diet (n = 60) Genotype-discordant diet (n = 62)

Mean (SD) Mean (SD) Adjusted differencea (95% CI)
b
Restraint (EI) 3.6 (0.9) 3.3 (1.0) 0.4 (−1.1, 1.9)
Disinhibition (EI)c −0.1 (0.6) 0.1 (0.7) 0.0 (−1.0, 0.9)
Hunger (EI)d −0.4 (0.5) −0.9 (0.6) 0.5 (−0.4, 1.4)
HF/HS (FPQ) −0.1 (0.3) 0.0 (0.4) 0.0 (−0.5, 0.5)
LF/HS (FPQ) 0.1 (0.3) 0.1 (0.3) 0.1 (−0.4, 0.5)
HF/HCCHO (FPQ) −0.3 (0.3) −0.3 (0.3) 0.0 (−0.4, 0.5)
LF/HCCHO (FPQ) −0.1 (0.3) 0.0 (0.3) −0.1 (−0.5, 0.4)
HF/LCHO/HP (FPQ) −0.4 (0.3) −0.4 (0.3) 0.0 (−0.5, 0.4)
LF/LCHO/HP (FPQ) 0.1 (0.3) 0.1 (0.3) 0.0 (−0.4, 0.4)
Fat-responders High-fat diet (n = 44) High-carbohydrate diet (n = 41)
Mean (SD) Mean (SD) Adjusted differencea (95% CI)
Restraint (EI)b 3.5 (1.2) 2.7 (1.4) 0.8 (−1.3, 2.9)
Disinhibition (EI)c −0.3 (0.8) 0.2 (0.9) −0.4 (−1.6, 0.9)
Hunger (EI)d −0.9 (0.7) −1.3 (0.8) 0.4 (−0.8, 1.5)
HF/HS (FPQ) 0.0 (0.4) 0.0 (0.5) 0.1 (−0.6, 0.7)
LF/HS (FPQ) 0.2 (0.4) 0.3 (0.4) 0.0 (−0.6, 0.5)
HF/HCCHO (FPQ) −0.2 (0.4) −0.1 (0.4) 0.0 (−0.6, 0.5)
LF/HCCHO (FPQ) 0.0 (0.4) 0.2 (0.4) −0.2 (−0.7, 0.4)
HF/LCHO/HP (FPQ) −0.5 (0.4) −0.6 (0.4) 0.1 (−0.5, 0.7)
LF/LCHO/HP (FPQ) 0.2 (0.4) 0.1 (0.4) 0.1 (−0.4, 0.7)
Carbohydrate- High-carbohydrate diet (n = 16) High-fat diet (n = 21)
responders Mean (SD) Mean (SD) Adjusted differencea (95% CI)
Restraint (EI)b 3.4 (1.1) 4.6 (1.1) −0.7 (−2.6, 1.2)
Disinhibition (EI)c 0.7 (0.9) 0.0 (0.9) 0.8 (−0.8, 2.5)
Hunger (EI)d 0.8 (0.8) −0.1 (0.9) 1.0 (−0.4, 2.5)
HF/HS (FPQ) −0.2 (0.5) 0.0 (0.5) −0.1 (−1.0, 0.7)
LF/HS (FPQ) −0.1 (0.4) −0.4 (0.4) 0.3 (−0.3, 0.9)
HF/HCCHO (FPQ) −0.4 (0.5) −0.6 (0.5) 0.2 (−0.6, 1.0)
LF/HCCHO (FPQ) −0.2 (0.4) −0.5 (0.4) 0.3 (−0.3, 0.9)
HF/LCHO/HP (FPQ) −0.3 (0.4) 0.1 (0.5) −0.2 (−1.0, 0.5)
LF/LCHO/HP (FPQ) −0.1 (0.4) 0.1 (0.4) −0.2 (−0.8, 0.4)
CI confidence interval, EI Eating Inventory, FPQ Food Preference Questionnaire, HF/HS high fat/high simple sugar, LF/HS low fat/high simple sugar, HF/HCCHO high fat/high complex carbo-
hydrate, LF/HCCHO low fat/high complex carbohydrate, HF/LCHO/HP high fat/low carbohydrate/high protein, LF/LCHO/HP low fat/low carbohydrate/high protein, SD standard deviation.
a
Adjusted for sex, race, and baseline value of the outcome.
b
Genotype-concordant diet: 46/60 participants; genotype-discordant diet: 47/62 participants. Fat-responders: 34/44 participants (high-fat diet) and 29/41 participants (high-carbohydrate diet).
Carbohydrate-responders: 12/16 participants (high-carbohydrate diet) and 18/21 participants (high-fat diet).
c
Genotype-concordant diet: 49/60 participants; genotype-discordant diet: 49/62 participants. Fat-responders: 37/44 participants (high-fat diet) and 31/41 participants (high-carbohydrate diet).
Carbohydrate-responders: 12/16 participants (high-carbohydrate diet) and 18/21 participants (high-fat diet).
d
Genotype-concordant diet: 51/60 participants; genotype-discordant diet: 51/62 participants. Fat-responders: 37/44 participants (high-fat diet) and 33/41 participants (high-carbohydrate diet).
Carbohydrate-responders: 14/16 participants (high-carbohydrate diet) and 18/21 participants (high-fat diet).

Pennington Biomedical Research Center (PBRC, Baton Rouge, LA). conditions, diseases, or medications that affect body weight or meta-
Participants were enrolled between October 7, 2020 and September 8, bolism or could affect risk or study completion, and a genotype indi-
2021. Participants were identified a priori as carbohydrate-responders cating a predisposition to respond favorably to neither or both of the
and fat-responders based on their combined genotypes at 10 genetic specified diets. We estimated that approximately 1/3 of people would
variant loci and randomized to either a high-carbohydrate or high-fat be fat-responders, 1/3 carbohydrate-responders, and 1/3 would
diet, yielding the following groups: (1) fat-responders receiving a high- respond favorably to neither or both of the specified diets.
fat diet, (2) fat-responders receiving a high-carbohydrate diet, (3) The study included 1 orientation visit, 2 clinic visits (one before
carbohydrate-responders receiving a high-fat diet, and (4) and one after the intervention), and 12-weekly intervention sessions.
carbohydrate-responders receiving a high-carbohydrate diet. All participants provided written informed consent, and participants
Participants were recruited from the community. Eligible partici- who completed the study received a minor compensation of $150.
pants were 18–75 years old, had a BMI of 27.0–47.5 kg/m2, and had
completed or were willing to complete a genealogy test (e.g., Ancestry, Genotype determination. Carbohydrate- and fat-responders were
23andMe) and to share the raw data with the investigators. Finally, a identified a priori based on their combined genotypes at the following
genetic profile indicating a predisposition to respond favorably to a genetic variants: (1) FGF21rs83814725, (2) TCF7L2rs1225537226,43, (3)
high-carbohydrate or high-fat WL diet based on specific SNPs (see IRS1rs294364128, (4) APOA5rs66279930,31,44, (5) PLIN1rs89416027,32, (6)
below) was required. Exclusion criteria included smoking, weight APOA2rs508229,33, (7) FTOrs993960934,35, (8) PPARGrs180128236, (9)
change ≥10 lbs. in the last 3 months, being pregnant or breastfeeding, GIPRrs1042392837, and (10) GYS2rs147829038. The genetic information

Nature Communications | (2023)14:6321 8

Article https://doi.org/10.1038/s41467-023-41969-1

Table 5 | Change in items of the Diet Personalization Survey during the 12-week intervention in those assigned to a diet
concordant vs. discordant with the genotype

All participants Genotype-concordant diet (n = 60) Genotype-discordant diet (n = 62)

Meana (SD) Meana (SD) Adj. differenceb (95% CI)
The assigned diet…
… fits my typical eating habits 0.9 (0.5) 1.3 (0.6) −0.3 (−1.2, 0.6)
… fits my lifestyle 0.4 (0.6) 0.2 (0.6) 0.2 (−0.7, 1.1)
… makes it easier to lose weight 0.6 (0.5) 0.7 (0.6) 0.1 (−0.8, 0.8)
I am confident that I can…
… successfully lose weight on the assigned diet 0.4 (0.5) 0.6 (0.5) −0.1 (−0.8, 0.7)
… follow the assigned diet −0.7 (0.4) −0.4 (0.5) −0.3 (−0.9, 0.4)
Fat-responders High-fat diet (n = 44) High-carbohydrate diet (n = 41)
Meana (SD) Meana (SD) Adj. differenceb (95% CI)
The assigned diet…
… fits my typical eating habits 0.6 (0.6) 1.3 (0.7) −0.6 (−1.6, 0.4)
… fits my lifestyle 0.0 (0.7) 0.2 (0.8) −0.1 (−1.2, 1.0)
… makes it easier to lose weight 0.2 (0.7) 0.4 (0.7) −0.1 (−1.1, 0.9)
I am confident that I can…
… successfully lose weight on the assigned diet 0.1 (0.6) 0.4 (0.7) −0.1 (−1.1, 0.8)
… follow the assigned diet −0.9 (0.6) −0.5 (0.6) −0.4 (−1.2, 0.5)
Carbohydrate-responders High-carbohydrate diet (n = 16) High-fat diet (n = 21)
Meana (SD) Meana (SD) Adj. differenceb (95% CI)
The assigned diet…
… fits my typical eating habits 1.6 (1.1) 1.3 (1.2) 0.4 (−1.4, 2.2)
… fits my lifestyle 1.1 (1.0) 0.3 (1.1) 1.0 (−0.7, 2.7)
… makes it easier to lose weight 1.3 (0.7) 1.3 (0.8) 0.4 (−0.8, 1.6)
I am confident that I can…
… successfully lose weight on the assigned diet 0.9 (0.8) 1.1 (0.9) 0.2 (−1.1, 1.5)
… follow the assigned diet −0.3 (0.7) −0.2 (0.7) 0.0 (−1.1, 1.1)
… the degree to which the diet helped manage 6.8 (0.9) 6.6 (1.0) 0.3 (−1.3, 1.8)
hunger
a
Mean change during the 12-week intervention.
b
Adjusted for sex and race.

was accessed via the raw data from the genealogy tests. Initially, only 6 The PBRC biostatistics department created the randomization
SNPs were included and pilot tested, and the scoring criteria were then sequence using SAS 9.4 statistical software for Windows (SAS Institute,
modified as few participants were deemed carbohydrate- or fat- Cary, NC) and uploaded it to REDCap (Research Electronic Data
responders. The original and updated scoring criteria, including a Capture). REDCap used strata for the inaction of genotype and gender.
specific example for 1 SNP, are provided in the Supplementary Meth- To ensure a relatively equal baseline BMI between the 4 genotype-diet
ods, including Supplementary Tables 1 and 2. The final risk score groups, a 1:1 randomization scheme was devised that adjusted for
comprised 10 SNPs with demonstrated and validated effects on the BMI, gender, and genotype. Gender and genotype were used as strata,
responses to high-fat/high-carbohydrate diets25–38,43,44, and validation while BMI was used in an a-priori-created randomization equation.
of this comprehensive and informative risk score was an objective of Within each stratum, this equation used block sizes of 6 (for females)
this study. and 4 (for males) at the start of the study and ended with block sizes
of 4 and 2, respectively, to ensure relative balance of group assign-
Intervention ments. Block sizes were assigned during the study by the biostatisti-
After enrollment (Week [W] 0 visit), participants were randomized to cian with access only to information about the enrolment progress
either a high-carbohydrate diet (rich in whole-grain foods) or a high-fat (percent enrolled).
diet (rich in unsaturated fats/oils). The high-carbohydrate diet con- Outcome assessors were blind to diet assignment and genotype
sisted of ~20% of energy from fat and ~65% from carbohydrates, patterns. Interventionists administering intervention sessions were
whereas the high-fat diet consisted of ~40% energy from fat and ~45% blind to genotype patterns but not diet type. Participants were only
from carbohydrates. Both diets provided 15% of energy from protein. informed of their genotype (carbohydrate- or fat-responder) once they
All participants were assigned an energy intake target that would result completed the study.
in a daily deficit of ~750 kcal and provided with a diet-specific meal plan The 12 weekly intervention (group) sessions were diet-specific and
in 200 kcal increments from 1400 to 2800 kcal/day to self-prepare had a different focus each week (Supplementary Material). Participants
meals during the intervention period. To facilitate meal plan adher- were provided a body weight scale and encouraged to weigh daily
ence when preparing or selecting meals, the meal plans included a list throughout the intervention and to send pictures of their weights to
of ingredients (and their amounts) for all meals of each day (breakfast, their interventionist before each intervention session. With very few
lunch, dinner, and 1 daily snack) and instructions for meal preparation exceptions, the first intervention session was conducted in person.
and participants were provided a food scale. Baseline energy require- Due to the COVID-19 pandemic, almost all subsequent sessions were
ments were calculated with Mifflin-St. Jeor’s formulas45. conducted virtually via webinar (Microsoft Teams).

Nature Communications | (2023)14:6321 9

Article https://doi.org/10.1038/s41467-023-41969-1

Table 6 | Change in intervention satisfaction (post-intervention) in those assigned to a diet concordant vs. discordant with the
genotype

All participants Genotype-concordant diet (n = 60) Genotype-discordant diet (n = 62)

Meana (SD) Meana (SD) Adj. differenceb (95% CI)
I am satisfied with…
… the group format 6.9 (0.4) 7.4 (0.4) −0.5 (−1.0, 0.1)
… the support from interventionists 7.5 (0.3) 7.5 (0.3) 0.1 (−0.4, 0.5)
… the intervention materials 7.0 (0.3) 7.2 (0.3) −0.1 (−0.6, 0.4)
… the support from other participants 6.4 (0.4) 6.5 (0.4) −0.1 (−0.7, 0.6)
… the amount of food in my meal plan 6.5 (0.5) 6.4 (0.5) 0.1 ( − 0.6, 0.8)
… the macronutrient content in my meal plan 6.1 (0.4) 5.8 (0.5) 0.3 (−0.4, 1.0)
… my progress toward weight management 6.4 (0.5) 6.3 (0.5) 0.3 (−0.5, 1.0)
… the degree to which the diet helped manage hunger 6.5 (0.5) 6.1 (0.5) 0.5 (−0.3, 1.2)
Fat-responders High-fat diet (n = 44) High-carbohydrate diet (n = 41)
Meana (SD) Meana (SD) Adj. differenceb (95% CI)
I am satisfied with…
… the group format 6.8 (0.4) 7.3 (0.5) −0.6 (−1.2, 0.1)
… the support from interventionists 7.6 (0.4) 7.8 (0.4) −0.1 (−0.7, 0.4)
… the intervention materials 7.1 (0.4) 7.4 (0.4) −0.3 (−0.9, 0.3)
… the support from other participants 6.1 (0.5) 6.5 (0.6) −0.4 (−1.2, 0.4)
… the amount of food in my meal plan 6.3 (0.6) 6.3 (0.6) 0.0 (−0.9, 0.9)
… the macronutrient content in my meal plan 6.0 (0.5) 5.6 (0.6) 0.3 (−0.5, 1.2)
… my progress toward weight management 6.5 (0.6) 6.2 (0.6) 0.4 (−0.5, 1.2)
… the degree to which the diet helped manage hunger 6.4 (0.6) 5.8 (0.6) 0.6 (−0.3, 1.4)
Carbohydrate-responders High-carbohydrate diet (n = 16) High-fat diet (n = 21)
Meana (SD) Meana (SD) Adj. differenceb (95% CI)
I am satisfied with…
… the group format 7.2 (0.7) 7.4 (0.7) −0.2 (−1.3, 0.9)
… the support from interventionists 7.5 (0.6) 6.9 (0.6) 0.5 (−0.4, 1.4)
… the intervention materials 7.1 (0.5) 6.6 (0.5) 0.4 (−03, 1.1)
… the support from other participants 7.1 (0.6) 6.4 (0.7) 0.8 (−0.3, 1.8)
… the amount of food in my meal plan 7.0 (0.7) 6.6 (0.8) 0.4 (−0.8, 1.6)
… the macronutrient content in my meal plan 6.3 (0.8) 5.9 (0.9) 0.4 (−1.0, 1.7)
… my progress toward weight management 6.2 (0.9) 6.4 (0.9) −0.1 (−1.5, 1.4)
… the degree to which the diet helped manage hunger 6.8 (0.9) 6.6 (1.0) 0.3 (−1.3, 1.8)
a
Mean post-intervention value. The Intervention Satisfaction Survey was only assessed at Week 12.
b
Adjusted for sex and race.

Outcome measures W12, as well as during the intervention session at W6, and the Inter-
Anthropometric data. At W0 and W12, fasting body weight and waist vention Satisfaction Survey (Supplementary Methods) was conducted
and hip circumference were measured in the PBRC outpatient clinic. at W12. Data for these surveys were collected and managed using
Clinic weights were also measured at all intervention visits (though not REDCap tools.
fasting weights). Further, body fat (%, via bioelectrical impedance
analysis [BIA]; X-contact 365, Jawon Medical Co., Ltd, Seoul, South Diet adherence. As stated above, participants were provided with a
Korea) and blood pressure (after 5 min of seated rest) were measured kitchen scale and could precisely weigh all ingredients specified
at W0 and W12. in the meal plans for the foods consumed at home. Additional
foods that were consumed were weighed and added as well. Adher-
Fasting serum glucose and insulin. Fasting serum glucose and insulin ence to the macronutrient content of the assigned diets was assessed
were measured at W0, and HOMA-IR was used to quantify insulin for three 7-day periods throughout the intervention (W4, W8, W12).
resistance.
Statistical analyses
Appetitive traits, food cravings, and food preferences. Appetitive The distribution of variables was evaluated by visual examination and
traits were measured with the Eating Inventory (EI)46, food cravings the Shapiro-Wilk test. The primary outcome was weight change (kg) at
were measured with the Food Craving Inventory (FCI)24, and hedonic 12 weeks. All other measures were secondary endpoints. Changes in
food preferences were measured with the Food Preference Ques- outcomes are presented as mean and 95% confidence intervals (CI). We
tionnaire (FPQ)47 at W0 and W12 (see Supplementary Methods for used linear mixed models to determine if changes in outcome vari-
details on outcome materials). Data for these questionnaires were ables differed among diets. Covariates in the models included baseline
collected and managed using REDCap tools value of the outcome, sex, and race. The mixed-effect model accoun-
ted for the correlation of the subject over time, and least-square means
Diet personalization and intervention satisfaction. The Diet Perso- based on the estimate from the mixed-effect model were used to test
nalization Survey (Supplementary Methods) was completed at W0 and for differences in weight change between diets. To evaluate whether

Nature Communications | (2023)14:6321 10

Article https://doi.org/10.1038/s41467-023-41969-1

Fig. 4 | Adherence to the macronutrient compositions of the respective diet at week 4, n = 38 at week 8, and n = 37 at week 12), they were 45% carbohydrates (b),
week 4, week 8, and week 12. Boxplots showing adherence data for the high- 40% fat (d), and 15% protein (f). The dashed line shows the target intake with the
carbohydrate diet (a, c, e) and the high-fat diet (b, d, f). For the high-carbohydrate shaded area representing ±5%. In the boxplots, the center line denotes the median
diet (n = 22 at week 4 and 8 and n = 21 at week 12), target intakes were 65% value (50th percentile), the bounds of the box represent the 25th and 75th
carbohydrates (a), 20% fat (c), and 15% protein (e) and for the high-fat diet (n = 40 at percentiles of the dataset, and the whiskers mark the 5th and 95th percentiles.

baseline insulin levels and HOMA-IR needed to be included as covari- lose more weight than those on a genotype-discordant diet. Based on
ates, their effects on WL were tested using a linear mixed model, previous studies49,50, we assumed a standard deviation for between-
adjusted for diet group and other known covariates. Neither baseline group differences in weight change of 2.8 kg. To detect a 2.0 kg differ-
insulin levels nor HOMA-IR was significantly associated with WL; hence ence in weight change between group 1 (fat-responders on a high-fat
these variables were not included as covariates. The significance level diet) and group 2 (fat-responders on a high-carbohydrate diet) or
was set to 0.05 (2-sided). Multiple testing adjustment was performed between group 3 (carbohydrate-responders on a high-fat diet) and
for secondary outcomes using the Holm-Bonferroni method48. All group 4 (carbohydrate-responders on a high-carbohydrate diet), with
analyses were conducted using SAS (Windows version 9.4; SAS Insti- the intended sample size and an alpha level of 0.05, the present study
tute, Cary, NC) and the statistical program R version 4.0.2 (https://cran. would have 80% power. Further, based on the same assumptions, the
r-project.org/). present study would have >95% power to test if WL differs between
participants on a genotype-concordant diet (groups 1 and 4 combined)
Power calculations. The present study planned to obtain data from up and those on a genotype-discordant diet (groups 2 and 3).
to 154 participants in total, and we aimed to complete 32 participants per
genotype-diet group (128 participants in total) though we did not limit Reporting summary
recruitment to achieve equal numbers of participants in each group. We Further information on research design is available in the Nature
hypothesized that participants on a genotype-concordant diet would Portfolio Reporting Summary linked to this article.

Nature Communications | (2023)14:6321 11

Article https://doi.org/10.1038/s41467-023-41969-1

Data availability 18. Unick, J. L., Pellegrini, C. A., Demos, K. E. & Dorfman, L. Initial weight
All of the data needed to recapitulate the analysis found within this loss response as an indicator for providing early rescue efforts to
study can be found in the manuscript, figures and supplementary improve long-term treatment outcomes. Curr. Diab. Rep. 17,
information. Source data are provided with this paper. Due to privacy 69 (2017).
reasons, de-identified data from the study cannot be shared publicly 19. Bazzano, L. A. et al. Effects of low-carbohydrate and low-fat diets.
but will be available from the corresponding author (chris- Ann. Intern. Med. 161, 309–318 (2014).
[email protected]) immediately following the publication of 20. Ebbeling, C. B., Leidig, M. M., Feldman, H. A., Lovesky, M. M. &
the paper upon reasonable request. The study protocol and statistical Ludwig, D. S. Effects of a low–glycemic load vs low-fat diet in
analysis plan will also be available. Source data are provided with obese young adults: a randomized trial. JAMA 297, 2092–2102
this paper. (2007).
21. Gardner, C. D., Offringa, L. C., Hartle, J. C., Kapphahn, K. & Cherin, R.
References Weight loss on low-fat vs. low-carbohydrate diets by insulin resis-
1. Fryar, C. D., Carroll, M. D. & Afful, J. Prevalence of overweight, tance status among overweight adults and adults with obesity: a
obesity, and severe obesity among adults aged 20 and over: United randomized pilot trial. Obesity 24, 79–86 (2016).
States, 1960–1962 through 2017–2018. (2020). 22. Martin, C. K. et al. Change in food cravings, food preferences, and
2. Kopelman, P. Health risks associated with overweight and obesity. appetite during a low-carbohydrate and low-fat diet. Obesity 19,
Obes. Rev. 8, 13–17 (2007). 1963–1970 (2011).
3. Tremmel, M., Gerdtham, U.-G., Nilsson, P. M. & Saha, S. Economic 23. Myers, C. A., Martin, C. K. & Apolzan, J. W. Food cravings and body
burden of obesity: a systematic literature review. Int. J. Environ. Res. weight: a conditioning response. Curr. Opin. Endocrinol. Diabetes
Public. Health. 14, 435 (2017). Obes. 25, 298–302 (2018).
4. Hruby, A. & Hu, F. B. The epidemiology of obesity: a big picture. 24. White, M. A., Whisenhunt, B. L., Williamson, D. A., Greenway, F. L. &
Pharmacoeconomics 33, 673–689 (2015). Netemeyer, R. G. Development and validation of the food-craving
5. Shai, I. et al. Weight loss with a low-carbohydrate, Mediterranean, or inventory. Obes. Res. 10, 107–114 (2002).
low-fat diet. N. Engl. J. Med. 359, 229–241 (2008). 25. Heianza, Y. et al. Macronutrient intake-associated FGF21 genotype
6. Sacks, F. M. et al. Comparison of weight-loss diets with different modifies effects of weight-loss diets on 2-year changes of central
compositions of fat, protein, and carbohydrates. N. Engl. J. Med. adiposity and body composition: The POUNDS Lost Trial. Diabetes
360, 859–873 (2009). Care 39, 1909–1914 (2016).
7. Johnston, B. C. et al. Comparison of weight loss among named diet 26. Grau, K. et al. TCF7L2 rs7903146-macronutrient interaction in obese
programs in overweight and obese adults: a meta-analysis. JAMA. individuals’ responses to a 10-wk randomized hypoenergetic diet.
312, 923–933 (2014). Am. J. Clin. Nutr. 91, 472–479 (2010).
8. Dopler Nelson, M., Prabakar, P., Kondragunta, V., Kornman, K. & 27. Smith, C. E. et al. Perilipin polymorphism interacts with dietary
Gardner, C. Genetic phenotypes predict weight loss success: the carbohydrates to modulate anthropometric traits in hispanics of
right diet does matter: Paper presented at: joint conference of the Caribbean origin. J. Nutr. 138, 1852–1858 (2008).
50th Cardiovascular Disease Epidemiology and Prevention and 28. Qi, Q. et al. Insulin receptor substrate 1 gene variation modifies
Nutrition, Physical Activity, and Metabolism; March 2–3, 2010, San insulin resistance response to weight-loss diets in a 2-year rando-
Francisco, CA. mized trial: the Preventing Overweight Using Novel Dietary Strate-
9. Gardner, C. D. et al. Effect of low-fat vs low-carbohydrate diet on 12- gies (POUNDS LOST) trial. Circulation 124, 563–571 (2011).
month weight loss in overweight adults and the association with 29. Corella, D. et al. APOA2, dietary fat, and body mass index: replica-
genotype pattern or insulin secretion: The DIETFITS Randomized tion of a gene-diet interaction in 3 independent populations. Arch.
Clinical Trial. JAMA. 319, 667–679 (2018). Intern. Med. 169, 1897–1906 (2009).
10. Qi, L., Bray, G. A. & Sacks, F. M. Low-fat vs low-carbohydrate diets 30. Sánchez-Moreno, C. et al. APOA5 gene variation interacts with
and weight loss. JAMA. 320, 202–203 (2018). dietary fat intake to modulate obesity and circulating triglycerides
11. Bayer, S., Winkler, V., Hauner, H. & Holzapfel, C. Associations in a Mediterranean population. J. Nutr. 141, 380–385 (2011).
between genotype–diet interactions and weight loss—a systematic 31. Domínguez-Reyes, T. et al. Interaction of dietary fat intake with
review. Nutrients. 12, 2891 (2020). APOA2, APOA5 and LEPR polymorphisms and its relationship with
12. Cornier, M.-A. et al. Insulin sensitivity determines the effectiveness obesity and dyslipidemia in young subjects. Lipids Health Dis. 14,
of dietary macronutrient composition on weight loss in obese 106 (2015).
women. Obes. Res. 13, 703–709 (2005). 32. Holzbach, L. C., Silveira, A. G. Z., Franco, L. P., Horst, M. A. &
13. McClain, A. D., Otten, J. J., Hekler, E. B. & Gardner, C. D. Adherence Cominetti, C. Polymorphism PLIN1 11482 G>A interacts with dietary
to a low-fat vs. low-carbohydrate diet differs by insulin resistance intake to modulate anthropometric measures and lipid profile in
status. Diabetes Obes. Metab. 15, 87–90 (2013). adults with normal-weight obesity syndrome. Br. J. Nutr. 128,
14. Pittas, A. G. et al. A low-glycemic load diet facilitates greater weight 1004–1012 (2022).
loss in overweight adults with high insulin secretion but not in 33. Corella, D. et al. Association between the APOA2 promoter poly-
overweight adults with low insulin secretion in the CALERIE Trial. morphism and body-weight in Mediterranean and Asian popula-
Diabetes Care 28, 2939–2941 (2005). tions. Replication of a gene-saturated fat interaction. Int. J. Obes.
15. Martin, C. K. et al. Challenges in defining successful adherence to 35, 666–675 (2011).
calorie restriction goals in humans: Results from CALERIETM 2. Exp. 34. de Luis, D. A. et al. The rs9939609 gene variant in FTO modified the
Gerontol. 162, 111757 (2022). metabolic response of weight loss after a 3-month intervention with
16. Ponzo, V. et al. Predictors of attrition from a weight loss program. A a hypocaloric diet. J. Investig. Med. 61, 22–26 (2013).
study of adult patients with obesity in a community setting. Eat. 35. Xiang, L. et al. FTO genotype and weight loss in diet and lifestyle
Weight Disord. 26, 1729–1736 (2021). interventions: a systematic review and meta-analysis. Am. J. Clin.
17. Unick, J. L. et al. Weight change in the first 2 months of a lifestyle Nutr. 103, 1162–1170 (2016).
intervention predicts weight changes 8 years later. Obesity. 23, 36. Garaulet, M., Smith, C. E., Hernández-González, T., Lee, Y.-C. &
1353–1356 (2015). Ordovás, J. M. PPARγ Pro12Ala interacts with fat intake for obesity

Nature Communications | (2023)14:6321 12

Article https://doi.org/10.1038/s41467-023-41969-1

and weight loss in a behavioural treatment based on the Medi- was funded by WW International, Inc. (New York, NY, USA). The
terranean diet. Mol. Nutr. Food Res. 55, 1771–1779 (2011). concept and design of the study were led by the study PI (C.K.M.)
37. Qi, Q., Bray, G. A., Hu, F. B., Sacks, F. M. & Qi, L. Weight-loss diets and developed in collaboration with five other authors, one of
modify glucose-dependent insulinotropic polypeptide receptor whom is an employee of the sponsor. The sponsor had no role in
rs2287019 genotype effects on changes in body weight, fasting the execution of the study or statistical analysis. Two sponsor
glucose, and insulin resistance: the Preventing Overweight Using employees are co-authors of the paper and provided editorial
Novel Dietary Strategies trial. Am. J. Clin. Nutr. 95, 506–513 (2012). comments to the manuscript.
38. Seip, R. L. et al. Physiogenomic comparison of human fat loss in
response to diets restrictive of carbohydrate or fat. Nutr. Metab. 5, Author contributions
4 (2008). C.K.M. obtained funding for the study. C.K.M., C.H., J.W.A., J.L.D., J.M.O.,
39. Nielsen, D. E. & El-Sohemy, A. Disclosure of genetic information and C.M.C., M.I.C. and G.D.F. designed the study. C.K.M. and C.H. oversaw
change in dietary intake: a randomized controlled trial. PLoS ONE 9, data acquisition, and S.Y., C.H. and C.K.M. analyzed and interpreted the
e112665 (2014). data. C.H. and C.K.M. drafted the manuscript; all authors provided cri-
40. Celis-Morales, C., Lara, J. & Mathers, J. C. Personalising nutritional tical revisions for important intellectual content. C.K.M. was responsible
guidance for more effective behaviour change. Proc. Nutr. Soc. 74, for the overall study supervision, and F.L.G. was responsible for the
130–138 (2015). medical supervision.
41. O’Donovan, C. B., Walsh, M. C., Gibney, M. J., Brennan, L. & Gibney,
E. R. Knowing your genes: does this impact behaviour change? Funding
Proc. Nutr. Soc. 76, 182–191 (2017). Open Access funding enabled and organized by Projekt DEAL.
42. Arkadianos, I. et al. Improved weight management using genetic
information to personalize a calorie controlled diet. Nutr. J. 6, Competing interests
29 (2007). G.D.F. and M.I.C. are shareholders and employees at WW International,
43. Mattei, J., Qi, Q., Hu, F. B., Sacks, F. M. & Qi, L. TCF7L2 genetic Inc. (New York, NY, USA). C.K.M. has previously consulted for WW on a
variants modulate the effect of dietary fat intake on changes in body fee-for-service basis, with the latest consultation occurring in 2018. All
composition during a weight-loss intervention. Am. J. Clin. Nutr. 96, other authors declare no competing interests.
1129–1136 (2012).
44. Corella, D. et al. APOA5 gene variation modulates the effects of Additional information
dietary fat intake on body mass index and obesity risk in the Fra- Supplementary information The online version contains
mingham Heart Study. J. Mol. Med. 85, 119–128 (2007). supplementary material available at
45. Mifflin, M. D. et al. A new predictive equation for resting energy https://doi.org/10.1038/s41467-023-41969-1.
expenditure in healthy individuals. Am. J. Clin. Nutr. 51,
241–247 (1990). Correspondence and requests for materials should be addressed to
46. Stunkard, A. J. & Messick, S. The three-factor eating questionnaire Christoph Höchsmann.
to measure dietary restraint, disinhibition and hunger. J. Psycho-
som. Res. 29, 71–83 (1985). Peer review information Nature Communications thanks Jennie Brand-
47. Geiselman, P. J. et al. Reliability and validity of a macronutrient self- Miller, Christopher Gardner and the other, anonymous, reviewer(s) for
selection paradigm and a food preference questionnaire. Physiol. their contribution to the peer review of this work. A peer review file is
Behav. 63, 919–928 (1998). available.
48. Eichstaedt, K. E., Kovatch, K. & Maroof, D. A. A less conservative
method to adjust for familywise error rate in neuropsychological Reprints and permissions information is available at
research: the Holm’s sequential Bonferroni procedure. NeuroReh- http://www.nature.com/reprints
abilitation 32, 693–696 (2013).
49. Martin, C. K. et al. Efficacy of SmartLossSM, a smartphone-based Publisher’s note Springer Nature remains neutral with regard to jur-
weight loss intervention: results from a randomized controlled trial. isdictional claims in published maps and institutional affiliations.
Obesity 23, 935–942 (2015).
50. Heilbronn, L. K. et al. Effect of 6-month calorie restriction on bio- Open Access This article is licensed under a Creative Commons
markers of longevity, metabolic adaptation, and oxidative stress in Attribution 4.0 International License, which permits use, sharing,
overweight individuals: a randomized controlled trial. JAMA 295, adaptation, distribution and reproduction in any medium or format, as
1539–1548 (2006). long as you give appropriate credit to the original author(s) and the
source, provide a link to the Creative Commons licence, and indicate if
Acknowledgements changes were made. The images or other third party material in this
The center where the research was conducted is supported in part article are included in the article’s Creative Commons licence, unless
by U54 GM104940 from the National Institute of General Medical indicated otherwise in a credit line to the material. If material is not
Sciences of the National Institutes of Health, which funds the included in the article’s Creative Commons licence and your intended
Louisiana Clinical and Translational Science Center and by the use is not permitted by statutory regulation or exceeds the permitted
Nutrition Obesity Research Centers grant P30DK072476 titled use, you will need to obtain permission directly from the copyright
“Nutrition and Metabolic Health Through the Lifespan” sponsored holder. To view a copy of this licence, visit http://creativecommons.org/
by National Institute of Diabetes and Digestive and Kidney Diseases licenses/by/4.0/.
(NIDDK). C.H. was supported by an NIDDK National Research Ser-
vice Award (T32DK064584), and J.L.D. was funded by the American © The Author(s) 2023
Heart Association (Grant # 20POST35210907). The present study

Nature Communications | (2023)14:6321 13