Content uploaded by Chia-Wei Cheng
Author content
All content in this area was uploaded by Chia-Wei Cheng on Oct 20, 2017
Content may be subject to copyright.
METABOLIC DISEASE 2017 © The Authors,
some rights reserved;
exclusive licensee
American Association
for the Advancement
of Science.
Fasting-mimicking diet and markers/risk factors for aging,
diabetes, cancer, and cardiovascular disease
Min Wei,
1
* Sebastian Brandhorst,
1
* Mahshid Shelehchi,
1
Hamed Mirzaei,
1
Chia Wei Cheng,
1
Julia Budniak,
1
Susan Groshen,
2
Wendy J. Mack,
2
Esra Guen,
1
Stefano Di Biase,
1
Pinchas Cohen,
1
Todd E. Morgan,
1
Tanya Dorff,
3
Kurt Hong,
4
Andreas Michalsen,
5
Alessandro Laviano,
6
Valter D. Longo
1,7†
Calorie restriction or changes in dietary composition can enhance healthy aging, but the inability of most
subjects to adhere to chronic and extreme diets, as well as potentially adverse effects, limits their application.
We randomized 100 generally healthy participants from the United States into two study arms and tested the
effects of a fasting-mimicking diet (FMD)—low in calories, sugars, and protein but high in unsaturated fats—on
markers/risk factors associated with aging and age-related diseases. We compared subjects who followed 3 months
of an unrestricted diet to subjects who consumed the FMD for 5 consecutive days per month for 3 months. Three
FMD cycles reduced body weight, trunk, and total body fat; lowered blood pressure; and decreased insulin-like
growth factor 1 (IGF-1). No serious adverse effects were reported. After 3 months, control diet subjects were crossed
over to the FMD program, resulting in a total of 71 subjects completing three FMD cycles. A post hoc analysis of
subjects from both FMD arms showed that body mass index, blood pressure, fasting glucose, IGF-1, triglycerides,
total and low-density lipoprotein cholesterol, and C-reactive protein were more beneficially affected in participants
at risk for disease than in subjects who were not at risk. Thus, cycles of a 5-day FMD are safe, feasible, and effective
in reducing markers/risk factors for aging and age-related diseases. Larger studies in patients with diagnosed dis-
eases or selected on the basis of risk factors are warranted to confirm the effect of the FMD on disease prevention
and treatment.
INTRODUCTION
Metabolic syndrome is defined by co-occurrence of three of five of
the following conditions: abdominal obesity, elevated fasting glucose,
elevated blood pressure, high serum triglycerides, and low levels of
high-density lipoprotein (HDL) cholesterol (1). Affecting 47 million
Americans (2), it is associated with a major increase in the risk of
cardiovascular disease (CVD) and all-cause mortality (3). Although
prolonged fasting or very low calorie fasting-mimicking diets (FMDs)
can ameliorate the incidence of diseases such as cancer and multiple
sclerosis in mice (4–6), randomized trials to assess fasting’sabilityto
reduce markers/risk factors for aging and major age-related diseases
have not been carried out (7–9). Prolonged fasting, in which only water
is consumed for 2 or more days, reduces pro-growth signaling and acti-
vates cellular protection mechanisms in organisms ranging from single-cell
yeast to mammals (10). In mammals, this is achieved in part by tempora-
rily reducing glucose and circulating insulin-like growth factor 1 (IGF-1),
a hormone well studied for its role in metabolism, growth, and develop-
ment, as well as for its association with aging and cancer (11–16). Severe
growth hormone receptor and IGF-1 deficiencies are associated with a
reduced risk of cancer, diabetes, and overall mortality in humans (17,18).
Mice fed periodically with the FMD show extended healthspan and
multisystem regeneration, reduced inflammation and cancer inci-
dence, and enhanced cognitive performance (5). Despite its potential
for disease prevention and treatment, prolonged fasting is difficult to
implement in human subjects and may exacerbate preexisting nutri-
tional deficiencies, making it not feasible and/or safe for children, the
elderly, frail individuals, and even most of the healthy adults. We have
investigated whether a dietary intervention more practical and safer
than fasting could affect markers or risk factors for aging and diseases.
To this end, we developed an FMD based on a diet previously tested
in animals and designed to achieve effects similar to those caused
by fasting on IGF-1, insulin-like growth factor–binding protein
1 (IGFBP-1), glucose, and ketone bodies (17). To prevent nutrient
deficiency, this FMD provided between 3000 and 4600 kJ per day, as
well as high micronutrient nourishment, to each human subject (5).
We also previously showed the safety and feasibility of this interven-
tion in 19 study participants who consumed three monthly cycles of
this FMD lasting 5 days each (5).
We now report the results of a randomized controlled trial of 100
subjects, 71 of whom completed three cycles of the FMD either in a
randomized phase (n=39)orafterbeingcrossed over from a control
diet group to the FMD group (n=32).Weevaluatedtheeffectsofthe
FMD on risk factors and markers for aging, cancer, metabolic syn-
drome, and CVDs in generally healthy participants ranging from 20
to 70 years of age.
RESULTS
Baseline data for all subjects
From April 2013 to July 2015, 100 study participants were randomized
and assigned to either arm 1 (n= 48) or arm 2 (n= 52). At enrollment,
independent of whether they completed the trial or not, subjects in the
two arms were comparable for age, sex, race, and body weight (Table 1).
Hispanics (27%) were underrepresented in the study population in
1
Longevity Institute, School of Gerontology, and Department of Biological Sciences,
University of Southern California, Los Angeles, CA 90089, USA.
2
Department of Preventive
Medicine, Keck School of Medicine, University of Southern California, Los Angeles, CA
90033, USA.
3
Norris Comprehensive Cancer Center, Keck School of Medicine, University
of Southern California, Los Angeles, CA 90033, USA.
4
Department of Internal Medicine,
Keck School of Medicine, University of Southern California, Los Angeles, CA 90033, USA.
5
Department of Internal and Complementary Medicine, Charité University Medical
Center, 10117 Berlin, Germany.
6
Department of Clinical Medicine, Sapienza University,
00185 Rome, Italy.
7
FIRC Institute of Molecular Oncology, Italian Foundation for Cancer
Research Institute of Molecular Oncology, 20139 Milan, Italy.
*These authors contributed equally to this work.
†Corresponding author. Email: vlongo@usc.edu
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Wei et al., Sci. Transl. Med. 9, eaai8700 (2017) 15 February 2017 1of12
by guest on October 20, 2017http://stm.sciencemag.org/Downloaded from
comparison to their representation (~45%) in the greater Los Angeles
area (California) (19). The participants in control arm 1 were asked to
continue their normal diet for 3 months, whereas participants in arm 2
started the FMD intervention. Two participants withdrew from arm
1 because of scheduling conflicts before completion of the informed
consent. In the randomized comparison (Fig. 1), 18 participants or 5
of 48 (10%) in the control arm and 13 of 52 participants in the
FMD arm (25%) were excluded or withdrew from the study. Of the
48 subjects enrolled in the control arm, two withdrew because of schedul-
ing conflicts, two because of unspecified personal issues, and one for un-
known reasons. Six of the 52 subjects enrolled in the FMD arm withdrew
from the study because of scheduling conflicts, five withdrew because of
unspecified personal issues, and two participants were excluded from the
study because of noncompliance to the FMD protocol.
Adverse effects and safety
Following the Common Terminology Criteria for Adverse Events
(CTCAE; v4.0), 54 to 100% (depending on the adverse event) of the
participants reported no adverse effects during the FMD cycles (fig.
S1). The most common self-reported grade 1 (mild) or grade 2 (mod-
erate) symptoms experienced by the participants were fatigue, weak-
ness, and headaches. No adverse effects of grade 3 or higher were
reported. A comprehensive metabolic panel that measured changes
in metabolic markers and liver and kidney function showed no neg-
ative effects of three cycles of the FMD (table S1). In summary, after
three cycles of the FMD, subjects reported only some mild and very
few moderate side effects.
Baseline risk factors and metabolic markers: Comparison of
randomized control and FMD subjects who completed the trial
At baseline, there were no significant differences in metabolic markers
or risk factors for age-related diseases and conditions between the
subjects who successfully completed the randomized trial in arm
1 (normal diet) and arm 2 (FMD), including body weight (P=0.39),
body mass index (BMI) (P= 0.24), total body fat (P=0.11),trunkfat
(P= 0.087), lean body mass (P= 0.15), waist circumference (P=0.34),
fasting glucose (P=0.55),IGF-1(P= 0.51), systolic and diastolic blood
pressure (P=0.60andP= 0.91, respectively), triglycerides (P=0.21),
and C-reactive protein (CRP) (P= 0.28). The notable exception was
that total cholesterol (P= 0.014) and low-density lipoprotein (LDL)
(P= 0.024), but not HDL (P= 0.99), were significantly lower at baseline
for subjects who were enrolled and completed arm 2 (Table 2). In sum-
mary, the values for disease markers and risk factors at baseline were
comparable between the control diet and FMD groups, with the excep-
tion of total and LDL cholesterol.
Changes in risk factors and metabolic markers: Comparison
of randomized control and FMD groups
Next, we evaluated the effects of the FMD by assessing the changes
in marker/risk factor values between baseline and 5 to 7 days after
the end of the third cycle of the FMD and compared them to those
occurring in the control arm within the same 3-month period (Fig. 2,
Table2,andtableS2).ParticipantsintheFMDarm(arm2)loston
average 2.6 ± 2.5 kg (±SD) (P< 0.0001) of weight, which was due in
part to a reduction in total body fat (absolute values and relative vol-
ume % of total mass) and trunk fat (absolute values) (Table 2 and
table S2). Subjects on the control diet did not lose body weight (0.1 ±
2.1 kg). After controlling the false discovery rate (FDR) of 0.05 between
the control and FMD groups, no change in the percentage of lean body
mass was observed (relative to the total mass; P= 0.07), although ab-
solute lean body mass was reduced in arm 2 (P=0.004)(Table2and
table S2). Waist circumference measured after three FMD cycles was
reduced by 4.1 ± 5.2 cm (P= 0.0035 between groups). The FMD cycles
also resulted in a decrease in IGF-1 concentrations of 21.7 ± 46.2 ng/ml
(P= 0.0017 between groups). Systolic blood pressure was reduced by 4.5
±6.0mmHg(P=0.023 between groups), and diastolic blood pressure
was reduced by 3.1 ± 4.7 mmHg (P= 0.053 between groups). Fasting
glucose (P= 0.27), triglycerides (P= 0.27), cholesterol (total, P= 0.81;
Table 1. Characteristics of all subjects at enrollment. Plus-minus values
are means ± SD rounded to the nearest 10th.
Characteristics Arm 1 (n= 48) Arm 2 (n= 52)
Sex, n(%)
Male 18 (37.5) 19 (36.5)
Female 30 (62.5) 33 (63.5)
Race or ethnic group, n(%)*
White 26 (54.2) 25 (48.1)
Black 2 (4.2) 5 (9.6)
Hispanic 13 (27.1) 14 (26.9)
Asian 6 (12.5) 7 (13.5)
Other 1 (2.1) 1 (1.9)
Age (years) 42.2 ± 12.5 43.3 ± 11.7
Weight (kg) 77.0 ± 15.9 74.3 ± 16.6
Education (years) 16.7 ± 2.8 16.6 ± 2.3
Smoking status, n(%)
Never smoked 29 (60.4) 39 (75.0)
Former smoker 13 (27.1) 9 (17.3)
Current smoker 6 (12.5) 4 (7.7)
BMI, n(%)
†
Mean 27.8 ± 5.1 26.6 ± 4.9
<25 17 (35.4) 20 (38.4)
25–30 18 (37.5) 21 (40.4)
>30 13 (27.1) 11 (21.2)
Systolic blood pressure (mmHg) 117.2 ± 12.3 117.2 ± 13.0
Diastolic blood pressure (mmHg) 75.6 ± 9.2 75.2 ± 7.8
Triglycerides (mg/dl) 104.0 ± 64.6 84.7 ± 37.2
Cholesterol (mg/dl)
Total 197.5 ± 39.6 185.7 ± 36.6
LDL 114.5 ± 36.1 110.3 ± 61.6
HDL 62.2 ± 16.4 65.2 ± 18.1
*The race or ethnic group was assigned by the subjects themselves. †The
BMI is the weight in kilograms divided by the square of the height in meters.
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Wei et al., Sci. Transl. Med. 9, eaai8700 (2017) 15 February 2017 2of12
by guest on October 20, 2017http://stm.sciencemag.org/Downloaded from
LDL, P=0.50;HDL,P= 0.90), and the acute-phase inflammatory
marker CRP (P= 0.27) did not differ significantly between groups. A
graphical summary of these data is presented in Fig. 2. In conclusion,
three cycles of the FMD reduced body weight, trunk and total body
fat, blood pressure, and IGF-1 in comparison to a normal diet.
Changes in risk factors and metabolic markers of
age-related diseases and conditions: Observational
pre-post FMD comparison
After 3 months, 43 subjects from the control arm were crossed over
to the FMD intervention. Eleven (26%) of these subjects withdrew
before completing three FMD cycles (Fig. 1). Five of these participants
withdrew because of scheduling issues, and two subjects opted to leave
the trial for unspecificpersonal reasons. We also excluded four partic-
ipants based on nonadherence to the FMD protocol. The causes for
withdrawal/exclusion were comparable between the arms. Considering
both FMD treatment arms, 24 of the 95 participants (25%) were ex-
cluded or withdrew from the study before completion of the three
FMD cycles (arm 2, n=13FMDs;arm1,n= 11 after FMD crossover)
because of schedulingconflicts(total,n= 11: arm 2, n=6FMDs;arm
1, n= 5 after FMD crossover), personal issues (total, n=7:arm2,n=
5FMDs;arm1,n= 2 after FMD crossover), or dislike of the diet and/or
nonadherence to the dietary protocol (total, n=6:arm2,n=2FMD;
arm 1, n= 4 after FMD crossover) (Fig. 1). The 25% dropout rate for
participants during the FMD is higher than the 10% dropout rate ob-
served during control diet in arm 1, but this is expected consid-
ering that subjects in control diet group only dropped out because
of scheduling conflicts because they were allowed to remain on their
normal diet. Ninety-five (95%) subjects completed one cycle, and 71
(71%) subjects completed three cycles of the FMD. Compared to the
71 participants who completed the three FMD cycles in arms 1 and
2, the 24 subjects who dropped out were not different in age (42.5 ±
11.6 years versus 43.3 ± 13.1 years) or BMI (27.1 ± 4.9 versus 26.9 ±
4.7) but were mostly female (18% male versus 82% female; P= 0.0045,
Fisher’sexacttest;fig.S2).
Because the differential dropout rate during the FMD treatment
period (25% in FMD in the randomized arm 2 and/or after arm
1 crossover versus 10% in the randomized arm 1 control) may have
caused biases in estimates of the FMD treatment effect, we compared
the changes in trial outcomes between the two groups who completed
three FMD cycles (n=39FMDrandomizedarm2andn=32after
arm 1 crossover to FMD) using sensitivity analysis. Three FMD cycles
had comparable effects between subjects in arm 1 (after crossover) and
arm 2 (randomized) with the exception of HDL, which underwent
a greater reduction in arm 2 (P= 0.03), and the decrease in absolute
lean body mass, which was observed in arm 2 but not arm 1 (table
S2). Because the FMD had similar effects in both arms, we combined
the results from the two arms to assess the changes in metabolites
Fig. 1. CONSORT diagram. Consolidated Standards of Reporting Trials (CONSORT) diagram of 102 contacted subjects of which 100 were enrolled into the study two
arms. Arm 1 (n= 48), the “control”group, maintained their normal caloric intake for a 3-month monitoring period. Data were collected at enrollment and again after 3 months.
Participants in arm 2 (n= 52) started the FMD after randomization. The FMD is provided for 5 days permonthforthreeconsecutivecycles.Data were collected at enrollment,
at the completion of the first FMD cycle but before resuming normal dietary intake, and also on average 5 days after subjects resumed their normal diet after the final FMD
cycle. After the initial 3-month period, subjects in arm 1 also started the FMD. An optional follow-up visit in the clinic for analysis was offered to all participants about 3 months
after the completion of the third FMD cycle.
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Wei et al., Sci. Transl. Med. 9, eaai8700 (2017) 15 February 2017 3of12
by guest on October 20, 2017http://stm.sciencemag.org/Downloaded from
Table 2. Biomarker/risk factor changes in subjects who completed the trial. CI, confidence interval.
Variable n
Baseline* CTRL: 3 months after baseline Efficacy
(comparing D)
P
§
Mean ± SD (95% CI)
FMD: 5 days after third FMD cycle
Mean ± SD (95% CI) P
†
Difference: D
‡
Body weight (kg)
Control diet, arm 1 43 77.2 ± 16.5 (72.1–82.2) 77.3 ± 17.0 (72.0–82.5) 0.72 0.1 ± 2.1 <0.0001
||
FMD, arm 2 39 74.1 ± 15.5 (69.3–78.9) 71.6 ± 14.6 (67.0–76.1) <0.0001 −2.6 ± 2.5
BMI
¶
Control diet, arm 1 43 27.4 ± 4.8 (25.9–28.9) 27.4 ± 5.0 (25.9–28.9) 0.82 0.0 ± 0.7 <0.0001
||
FMD, arm 2 39 26.2 ± 4.4 (24.8–27.6) 25.3 ± 4.3 (24.0–26.5) <0.0001 −0.9 ± 0.9
Total body fat** (absolute volume)
Control diet, arm 1 43 23,651 ± 8,155 (21,142–26,161) 23,607 ± 8,337 (21,041–26,173) 0.83 −44 ± 1,365 0.0002
||
FMD, arm 2 38 20,643 ± 8,459 (17,953–23,332) 19,249 ± 7,792 (16,772–21,726) <0.0001 −1,393 ± 1,786
Trunk fat** (absolute volume)
Control diet, arm 1 43 8,429 ± 4,742 (6,969–9,888) 8,395 ± 4,776 (6,925–9,865) 0.83 −33 ± 1,046 0.018
FMD, arm 2 38 6,573 ± 4,877 (5,022–8,124) 5,938 ± 4,295 (4,572–7,303) 0.0023 −636 ± 1,198
Lean body mass** (relative volume %)
Control diet, arm 1 43 63.9 ± 8.2 (61.4–66.4) 64.0 ± 8.7 (61.3–66.7) 0.64 0.1 ± 1.5 0.070
FMD, arm 2 38 66.8 ± 9.6 (63.7–69.8) 67.6 ± 9.4 (64.6–70.6) 0.016 0.8 ± 2.0 −0.8 ± 25
Waist circumference (cm)
Control diet, arm 1 28 95.4 ± 14.2 (89.9–100.9) 94.6 ± 14.5 (88.9–100.2) 0.10 −0.8 ± 25 0.0035
||
FMD, arm 2 28 92.1 ± 11.2 (87.9–96.2) 87.9 ± 120 (83.5–92.4) 0.0003 −4.1 ± 5.2
Fasting glucose (mg/dl)
Control diet, arm 1 41 88.1 ± 8.9 (85.3–90.9) 90.3 ± 9.7 (87.3–93.4) 0.14 2.2 ± 9.5 0.27
FMD, arm 2 36 89.7 ± 8.5 (86.5–92.1) 89.0 ± 8.0 (86.4–91.6) 0.87 −0.8 ± 9.9
IGF-1 (ng/ml)
Control diet, arm 1 41 180.2 ± 84.5 (153.5–2,069) 188.9 ± 91.0 (160.2–217.7) 0.14 8.7 ± 36.9 0.0017
||
FMD, arm 2 38 168.6 ± 69.1 (146.6–190.5) 146.9 ± 62.3 (127.0–166.7) 0.0063 −21.7 ± 46.2
Systolic blood pressure (mmHg)
Control diet, arm 1 43 116.5 ± 12.3 (112.7–1,203) 115.8 ± 13.6 (111.6–120.0) 0.60 −0.7 ± 8.4 0.023
FMD, arm 2 38 118.0 ± 13.4 (113.7–1,222) 113.5 ± 13.2 (109.3–117.7) <0.0001 −4.5 ± 6.0
Diastolic blood pressure (mmHg)
Control diet, arm 1 43 75.5 ± 9.6 (72.5–78.5) 74.8 ± 10.0 (71.7–77.9) 0.45 −0.7 ± 6.2 0.053
FMD, arm 2 38 75.7 ± 8.0 (73.2–78.3) 72.6 ± 8.7 (70.5–76.0) 0.0089 −3.1 ± 4.7
Triglycerides (mg/dl)
Control diet, arm 1 37 100.5 ± 68.2 (77.7–123.2) 101.5 ± 57.1 (82.5–120.6) 0.85 1.0 ± 35.0 0.27
FMD, arm 2 30 83.0 ± 39.5 (69.1–96.9) 74.9 ± 37.6 (61.7–88.2) 0.19 −8.1 ± 33.5
continued on next page
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Wei et al., Sci. Transl. Med. 9, eaai8700 (2017) 15 February 2017 4of12
by guest on October 20, 2017http://stm.sciencemag.org/Downloaded from
and risk factors during the first FMD cycle (at day 5 of the FMD and
before refeeding; table S3) and after completion of three FMD cycles
(5 to 7 days after completing the third FMD cycle; table S2).
At the end of the first FMD cycle and before resuming the normal
diet, body weight (P< 0.0001), BMI (P< 0.0001), absolute lean body
mass (P< 0.0001), waist circumference (P< 0.0001), fasting glucose (P<
0.0001), IGF-1 (P< 0.0001), diastolic blood pressure (P< 0.0003), tri-
glycerides (P< 0.0001), and LDL (P< 0.0026) were significantly reduced
compared to baseline. In contrast, relative lean body mass (P=0.02),
b-hydroxybutyrate (P< 0.0001), and IGFBP-1 (P< 0.0001) were in-
creased. Both absolute and relative total body fat (P= 0.075 and P=
0.047, respectively), systolic blood pressure (P= 0.076), as well as CRP
(P= 0.75) were not significantly changed after completion of the first
FMD cycle compared to baseline (table S3). These results indicate that
subjects did follow the dietary changes imposed by the FMD and re-
sponded to them as anticipated.
In subjects who completed three FMD cycles (combining both
FMD arms) and who returned to the normal diet for 5 to 7 days, body
weight (P< 0.0001; n=71),BMI(P< 0.0001; n= 71), total body fat
(absolute, P< 0.0001; relative, P< 0.0001; n= 70), trunk fat (absolute,
P< 0.001; relative, P= 0.0002; n= 70), absolute lean body mass (P=
0.0001; n= 70), waist circumference (P< 0.0001; n= 52), IGF-1 (P<
0.0001; n= 69), systolic and diastolic blood pressure (P<0.0001and
P< 0.0004, respectively; n= 70), total cholesterol (P= 0.004; n= 55),
LDL (P< 0.0011; n=55),andHDL(
P=0
.02;n= 55) were signif-
icantly reduced, and relative lean body mass (P=0.0002;n= 70) was
increased. Fasting glucose (P=0.28;n=66),b-hydroxybutyrate (P=
0.23; n= 69), IGFBP-1 (P=0.84;n= 69), triglycerides (P=0.16;n=
55), and CRP (P= 0.052; n= 69) were not significantly changed
(table S2). In summary, the combined FMD groups from arms 1 and 2
confirmed that the FMD cycles promoted potent effects on many meta-
bolic markers and disease risk factors, which are maintained after
subjects return to their normal diet.
FMD effects stratified by baseline risk factor values: A post
hoc observational pre-post FMD comparison
Age-related physiological changes that lead to increased risk factors
occur before diseases can be diagnosed (20,21). We used the aggregated
FMD data of both study arms and performed a post hoc analysis of the
FMD effect on risk factors for CVD and metabolic syndrome, defined
as three of five of the following conditions: abdominal obesity, elevated
fasting glucose, elevated blood pressure, high serum triglycerides, and
low HDL cholesterol (1). We selected clinically relevant cutoffs and
compared normal and at-risk subjects for each risk factor: total choles-
terol>199mg/dlandLDLcholesterollevels >130 mg/dl are associated
with an increased risk for CVD (22), a fasting glucose >99 mg/dl indi-
cates impaired fasting glucose/prediabetes (23), and triglyceride
levels >100 mg/dl (24) as well as CRP >1 mg/liter are associated
with increased risk for CVD (25). For serum IGF-1, no clinically re-
levant risk level has been established, but a number of epidemiological
studies have associated IGF-1 levels above 200 ng/ml with various
cancers (17,26). We therefore compared the effect of FMD cycles
on subjects in the highest quartile of IGF-1 expression (>225 ng/ml)
with that on subjects with IGF-1 levels ≤225 ng/ml.
In a post hoc analysis, we tested how the changes in the FMD
normal and at-risk subgroups compared to those in the control diet
normal and at-risk subgroups, as defined by their baseline levels of var-
ious risk factors (Table 3). We observed a benefit of the FMD, but not in
the control arm, on BMI in all BMI subgroups (Pvalue for interaction =
0.03), although the FMD was particularly beneficial among subjects
Variable n
Baseline* CTRL: 3 months after baseline
Efficacy
(comparing D)
P
§
Mean ± SD (95% CI)
FMD: 5 days after third FMD cycle
Mean ± SD (95% CI) P
†
Difference: D
‡
Total cholesterol (mg/dl)
Control diet, arm 1 37 195.9 ± 38.9 (182.9–2,089) 183.9 ± 35.2 (172.1–195.6) 0.0015 −12.0 ± 21.3 0.81
FMD, arm 2 30 175.3 ± 25.3 (166.4–1,842) 164.4 ± 23.4 (156.1–172.6) 0.0012 −10.9 ± 17.0
LDL cholesterol (mg/dl)
Control diet, arm 1 37 111.2 ± 35.6 (99.4–123.1) 104.0 ± 31.8 (93.4–114.6) 0.018 −7.2 ± 17.7 0.50
FMD, arm 2 30 94.1 ± 23.0 (86.0–102.2) 89.7 ± 22.8 (81.7–97.7) 0.13 −4.4 ± 16.0
HDL cholesterol (mg/dl)
Control diet, arm 1 37 64.3 ± 16.1 (5.9.2–69.9) 59.3 ± 14.9 (54.3–64.3) 0.0002 −5.3 ± 7.8 0.90
FMD, arm 2 30 64.8 ± 17.2 (58.6–70.6) 59.6 ± 12.8 (55.1–64.2) 0.0097 −5.0 ± 10.0
C-reactive protein (mg/liter)
Control diet, arm 1 42 1.5 ± 1.9 (0.92–2.11) 1.9 ± 2.7 (1.07–2.75) 0.31 0.4 ± 2.5 0.27
FMD, arm 2 38 1.1 ± 1.3 (0.71–1.52) 1.0 ± 1.2 (0.61–1.37) 0.61 −0.1 ± 1.5
*No significant differences at baseline between arm 1 and arm 2 (FMD), with exception of total cholesterol (P= 0.014) and LDL (P= 0.024). †Pvalues
comparing within-group changes were calculated using paired two-tailed Student’sttest. ‡Plus-minus values are means ± SD rounded to the nearest
tenth. §Between-arm comparison was calculated using two-tailed two-sample equal variance ttests. Using the Benjamini-Hochberg method for controlling
the FDR of 0.05. ||Pvalues indicate that the null hypothesis of no difference between the control diet (arm 1) to FMD (arm 2) can be rejected. ¶The BMI
is the weight in kilograms divided by the square of the height in meters. **Analyzed by DEXA.
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Wei et al., Sci. Transl. Med. 9, eaai8700 (2017) 15 February 2017 5of12
by guest on October 20, 2017http://stm.sciencemag.org/Downloaded from
who were obese (BMI >30) at baseline. The FMD-dependent reduction
in IGF-1 was also larger in participants with baseline IGF-1 ≥225 ng/ml
(Pvalue for interaction = 0.018).
Next,weevaluatedtheeffectsize,that is, efficacy in normal and at-
risk subjects (Table 4) in subjects stratified by risk factor. Subjects with
a BMI of greater than 30 (obese) experienced a greater reduction in
BMI by the end of the three FMD cycles than those with a BMI of less
than 25 (P= 0.011 between groups) and BMI of 25 to 30 (P= 0.0011
between groups). Systolic blood pressure was reduced by 2.4 ± 6.3 mmHg
in subjects with baseline systolic blood pressure ≤120 mmHg but by
6.7 ± 6.9 mmHg in subjects with systolic blood pressure >120 mmHg
(P= 0.013 between groups), and diastolic blood pressure was re-
duced by 1.5 ± 5.1 mmHg in subjects with diastolic blood pressure
≤80 mmHg but by 5.5 ± 6.4 mmHg in those with baseline levels above
80 mmHg (P= 0.01 between groups). Fasting glucose did not change
for participants with baseline levels ≤99 mg/dl but was reduced by 11.8
± 6.9 mg/dl in participants with baseline fasting glucose >99 mg/dl (P<
0.0001 between groups); notably, this reduction brought glucose in
these subjects within the healthy range. IGF-1 levels in subjects with
baseline levels above 225 ng/ml were reduced by 55.1 ± 45.6 ng/ml,
nearly four times more (P< 0.001 between groups) than the 14.1 ±
39.9 ng/ml reduction observed in participants with IGF-1 concentrations
below 225 ng/ml. Triglyceride levels were reduced more in participants
with baseline levels >100 mg/dl (P= 0.0035 between groups). Total
cholesterol was reduced significantly more in participants with total
cholesterol higher than 199 mg/dl at baseline (P= 0.015 between
groups). LDL was reduced by 14.9 ± 21.7 mg/dl in those with total
cholesterol baseline levels above 199 mg/dl but was not reduced by
Fig. 2. Change analysis of metabolic variables during the randomization. Effects on aging/disease markers and risk factors in all subjects who completed the
randomized analysis in either the control arm or the FMD arm (5 to 7 days after the third cycle of FMD). (A) Body weight, (B) BMI, (C) total body fat, (D) trunk fat, (E) lean
body mass, (F) waist circumference, (G) serum glucose level, (H) insulin-like growth factor 1, (I) systolic blood pressure, (J) diastolic blood pressure, (K) triglycerides, (L)total
cholesterol, (M)LDL,(N)HDL,and(O) CRP were measured in both cohorts as described. The Dchange represents a comparison to baseline. All data are means ± SD. Between-
arm comparisons were calculated using two-tailed two-sample equal variance ttests. For some of the 100 enrolled participants, the nurses were unable to collect all the
samples/measurements from all subjects. We therefore excluded subjects with incomplete measurements from a particular marker group (see Table 2 for details). Abs,
absolute; Rel, relative; BP, blood pressure.
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Wei et al., Sci. Transl. Med. 9, eaai8700 (2017) 15 February 2017 6of12
by guest on October 20, 2017http://stm.sciencemag.org/Downloaded from
FMD cycles in normal-range subjects (P= 0.013 between groups).
There was no reduction (P= 0.094 between groups) in HDL levels
for those study participants with HDL levels below or above 50 mg/
dl at baseline. CRP was not reduced for subjects with levels below
1 mg/liter but was reduced by 1.6 ± 1.3 mg/liter and returned to
the normal levels in most subjects with baseline CRP higher than
1mg/liter(P= 0.0003 between groups). A graphical summary of these
data is presented in Fig. 3; before-after dot plots of individual subjects
in the control cohort as well as in normal and at-risk subjects in the
FMD cohort are presented in fig. S3.
ThisposthocanalysisindicatesthattheFMDhadmorepronounced
effects in at-risk participants than in those subjects with risk factor
values within the normal range, with the exception of HDL. Larger ran-
domized trials are necessary to confirm the results on the efficacy of the
FMD in the treatment of patients at risk for diseases.
Voluntary follow-up 3 months after FMD
We invited participants to return on a voluntary basis about 3 months
(actual mean follow-up time, 3.2 ± 1.3 months; n= 50) after their
third and final FMD cycle. In these subjects, the FMD’s effects on
body weight, BMI, waist circumference, glucose (in at-risk subjects),
IGF-1, and systolic (in at-risk subjects) and diastolic blood pressure
persisted for at least 3 months after the final FMD cycle (table S4).
Subjects with low HDL levels at baseline displayed increased HDL
levels at the 3-month follow-up, whereas CRP levels remained signif-
icantly lower in study participants with baseline CRP levels above
1 mg/liter. Notably, some of the at-risk groups include only a few sub-
jects, and thus, larger studies are needed to establish long-term effects
of the FMD on disease risk factors.
These results indicate that some of the beneficial effects of multiple
cycles of the FMD may last for several months. Although subjects were
not advised to change their diet or exercise regimen after the FMD
cycles ended, we cannot rule out that some of the changes after the ad-
ditional 3 months may be a result of lifestyle changes such as healthier
diets and/or improved physical activity after the completion of this trial.
DISCUSSION
This randomized phase 2 trial indicates that three cycles of a 5-day FMD
per month are feasible, safe, and effective in reducing body weight, waist
circumference and BMI, absolute total body and trunk fat, systolic blood
pressure, as well as IGF-1. Metabolic markers such as fasting glucose,
triglycerides, and CRP, as well as total, HDL, and LDL cholesterol, which
were within the normal range at baseline, were not significantly affected
in the randomized comparison after three FMD cycles. After 3 months,
subjects from the control arm were crossed over to the FMD interven-
tion. Our post hoc analysis of the aggregated data from all 71 subjects
who completed three FMD cycles confirmed the effects of the FMD on
trunk and total body fat, blood pressure, and IGF-1. A post hoc analysis
also allowed us to analyze subjects with elevated risk factors or metabolic
markers associated with metabolic syndrome and age-related diseases,
such as high BMI, blood pressure, fasting glucose, triglycerides, CRP,
cholesterol, and IGF-1. The FMD had more pronounced effects on all
these markers in at-risk participants than in those subjects who had risk
factor values within the normal range. Some of these metabolic markers,
namely, CRP, systolic/diastolic blood pressure, and serum lipids, have
been proposed as markers of biological aging (27). However, other mar-
kers affected by the FMD, including IGF-1 and glucose, have been
strongly implicated in aging and age-related diseases (5,18,28).
Study participants were instructed not to alter their lifestyle for
the duration of the trial and were allowed to consume food of their
choice during the normal diet periods, that is, subjects were not placed
on a prespecified or calorie-restricted diet. We observed changes that
were both positive (total cholesterol and LDL) and negative (HDL) in
arm 1 subjects during the control diet period, potentially explained by
Table 3. Post hoc comparisons for changes in risk factors for age-related
diseases and conditions by baseline subgroups.
Subgroup
Group
differences
(FMD −Control)
Mean (95% CI)
Within
subgroup
P
Interaction
P
BMI
<25 −0.6 (−1.2 to −0.05) 0.03 0.03
25–30 −0.8 (−1.4 to −0.3) 0.003
>30 −1.9 (−2.6 to −1.1) 0.0009
Systolic blood pressure (mmHg)
<120 −3.4 (−7.2 to 0.5) 0.086 0.80
≥120 −4.3 (−10.4 to 1.8) 0.17
Diastolic blood pressure (mmHg)
<80 −2.5 (−5.3 to 0.3) 0.08 0.87
≥80 −3.0 (−8.2 to 2.3) 0.26
Fasting glucose (mg/dl)
<99 −0.8 (−5.2 to 3.6) 0.72 0.12
≥99 −11.7 (−25.0 to 1.5) 0.08
IGF-1 (ng/ml)
<225 −18.7 (−38.6 to 1.2) 0.065 0.018
≥225 −70.9 (−109.3 to −32.6) 0.0004
Triglycerides (mg/dl)
<100 −4.6 (−24.1 to 15.0) 0.64 0.38
≥100 −19.1 (−45.8 to 7.6) 0.16
Cholesterol (mg/dl)
Total, <199 −1.8 (−12.6 to 9.0) 0.73 0.88
Total, ≥199 −0.2 (−18.2 to 17.7) 0.98
LDL, <199 total
cholesterol
1.0 (−8.8 to 10.8) 0.84 0.60
LDL, ≥199 total
Cholesterol
6.2 (−11.2 to 23.6) 0.48
HDL, <50 −1.2 (−9.0 to 0.5) 0.75 0.70
HDL, ≥50 0.5 (−4.2 to 5.2) 0.83
CRP (mg/liter)
<1 −0.4 (−1.5 to 8.7) 0.47 0.59
≥1−0.9 (−2.4 to 0.6) 0.24
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Wei et al., Sci. Transl. Med. 9, eaai8700 (2017) 15 February 2017 7of12
by guest on October 20, 2017http://stm.sciencemag.org/Downloaded from
dietary habit changes in anticipation for the FMD, despite no change
inweight,BMI,bodyfat,orleanmass.Similarly,thepersistenteffects
of the FMD observed 3 months after study completion may be
affected by changes in dietary habits and/or physical activity. The
composition of the diet tested in this trial was based on the FMD that
isknowntoextendhealthspaninmice.Similarlytothestudyinmice
(5), we expect the FMD effects to be mostly independent of an overall
caloric restriction, because both groups likely consumed similar levels
of calories per month: For example, estimating a 9200 kJ diet for each
of the 25 to 26 nonrestricted days and about 19,200 kJ kcal for the
Table 4. Post-hoc analysis of risk factors for age-related diseases and conditions in at-risk subjects who completed the FMD trial.
Variable
(at baseline) n
Baseline FMD: 5 days after third FMD Efficacy
(comparing D)P
‡
Mean ± SD (95% CI) Mean ± SD (95% CI) P*D
†
BMI
§
<25 27 22.4 ± 1.7 (21.68–23.03) 21.9 ± 1.6 (21.24–22.52) 0.0014 −0.5 ± 0.7 0.0020
||
25–30 30 27.1 ± 1.4 (26.58–27.60) 26.4 ± 1.6 (25.76–26.99) <0.0001 −0.7 ± 0.7 0.21
>30 14 34.4 ± 3.5 (32.37–36.36) 33.0 ± 3.5 (30.95–34.95) 0.0001 −1.4 ± 1.0 0.011
Systolic blood pressure (mmHg)
<120 49 110.3 ± 6.9 (108.5–112.4) 107.9 ± 7.7 (105. 7–110.1) 0.0089 −2.4 ± 6.3 0.013
>120 21 133.4 ± 9.0 (129.3–137.5) 126.7 ± 11.3 (121.5–131.8) 0.0002 −6.7 ± 6.9
Diastolic blood pressure (mmHg)
<80 53 71.8 ± 5.0 (70.43–73.21) 70.3 ± 5.9 (68.64–71.89) 0.032 −1.5 ± 5.1 0.010
>80 17 87.7 ± 6.9 (84.20–91.27) 82.2 ± 10.0 (77.09–87.38) 0.0026 −5.5 ± 6.4
Fasting glucose (mg/dl)
<99 53 87.7 ± 6.4 (85.97–89.50) 89.0 ± 8.2 (86. 75–91.25) 0.29 1.3 ± 8.6 <0.0001
>99 13 104.2 ± 4.4 (101.5–106.9) 92.4 ± 8.2 (87.42–97.35) <0.0001 −11.8 ± 6.9
IGF-1 (ng/ml)
<225 52 145.1 ± 38.4 (134.4–155.8) 131.0 ± 46.8 (118.0–144.0) 0.014 −14.1 ± 39.9 0.00088
>225 17 286.6 ± 47.0 (261.6–311.7) 231.5 ± 64.7 (197.0–265.9) 0.0002 −55.1 ± 45.6
Triglycerides (mg/dl)
<100 34 65.8 ± 18.4 (59.45–72.09) 69.5 ± 28.9 (59.58–79.45) 0.33 3.7 ± 22.3 0.0035
>100 21 149.3 ± 41.2 (127.9–160.8) 123.7 ± 52.7 (97.54 –148.0) 0.058 −25.6 ± 50.2
Cholesterol (mg/dl)
Total, <199 40 170.3 ± 18.3 (164.4–176.2) 164.2 ± 20.2 (157. 7–170.6) 0.016 −6.1 ± 15.3 0.015
Total, >199 15 228.1 ± 23.0 (215.4–240.9) 208.2 ± 26.3 (193.6–222.8) 0.0088 −19.9 ± 25.4
LDL, <199 total
cholesterol
40 91.4 ± 20.2 (85.0 –97.73) 88.8 ± 21.0 (82.13–95.38) 0.22 −2.6 ± 13.4 0.013
LDL, >199 total
cholesterol
15 141.1 ± 29.1 (125.0–157.2) 126.2 ± 24.7 (112.2–139.9) 0.018 −14.9 ± 21.7
HDL, <50 17 44.1 ± 4.1 (41.94–46.17) 44.2 ± 4.6 (41.87–46.6) 0.82 0.1 ± 3.1 0.094
HDL, >50 38 69.6 ± 14.1 (65.06–74.17) 65.6 ± 11.6 (61.81–69.32) 0.015 −4.0 ± 10.0
CRP (mg/liter)
<1 43 0.4 ± 0.3 (0.28–0.46) 0.6 ± 0.9 (0.27–0.85) 0.20 0.2 ± 1.0 0.0003
>1 26 3.3 ± 2.8 (2.22–4.44) 1.6 ± 1.2 (1.06–2.15) 0.0085 −1.7 ± 3.1
*Pvalues comparing within-group changes were calculated using paired two-tailed Student’sttest. †Plus-minus values are means ± SD rounded to the
nearest tenth. ‡Between-group comparison was calculated using two-tailed two-sample equal variance ttests. §The BMI is the weight in kilograms
divided by the square of the height in meters. ||One-way analysis of variance for the BMI groups.
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Wei et al., Sci. Transl. Med. 9, eaai8700 (2017) 15 February 2017 8of12
by guest on October 20, 2017http://stm.sciencemag.org/Downloaded from
5 days of FMD per month, the between-group difference in consumed
calories is expected to be about 10%. In addition, this difference may
be overestimated because it is likely that subjects have an elevated calorie
intake after the FMD period, as we have shown for mice (5). Day 1 of
the FMD supplies ~4600 kJ (11% protein, 46% fat, and 43% carbohy-
drate), whereas days 2 to 5 provide ~3000 kJ per day (9% protein, 44%
fat, and 47% carbohydrate); thus, fat and complex carbohydrates are
the major source of calories in the FMD.
Our studies in cells and mice indicate that both glucose and pro-
teins interfere with the protective and regenerative effects of fasting
(29). Because our previous data indicate that dietary composition
can be equally or more important than calorie restriction, it will be
important to test the effects of a similarly restricted diet that provides
higher proportions of carbohydrates and/or proteins. It remains to be
established whether part of the effects of FMD that we observed are
mediated by stem cell–based regeneration or rejuvenation, as indicated
by our mouse studies (5).
The FMD-induced reduction in serum glucose and IGF-1 is of
interest given their role in pro-aging signaling pathways and cancer
(17,30–33). In addition to a marker for insulin resistance and a
metabolicinputforcancercells,glucose is associated with cellular sen-
sitization to toxins and senescence (28,34,35). Growth hormone re-
ceptor deficiency, resulting in reduced IGF-1 levels, is associated with
a major reduction in pro-aging signaling, cancer, and diabetes in
humans (18). The observed reduction in IGF-1 in our study, but not
either after 6 months of intermittent energy restriction (IER) (36)or
after 6 years of 20% caloric restriction (37), is probably related to the
long-lasting effects of the low protein/amino acid content of the FMD
(average 5 days of FMD; 11.5% versus 21% IER or 24% long-term CR). In
fact, twenty eight vegans consuming a moderately protein-restricted (10%)
diet for about 5 years had reduced IGF-1 levels compared to a group that
consumed a chronic 20% calorie-restricted diet (37). We also previously
showed that reduced IGF-1 levels and reduced cancer risk were asso-
ciated with low protein consumption in participants of the National
Health and Nutrition Examination Survey cohort (17). Specific ingre-
dients, for example, high levels of unsaturated fats and micronutrients,
may also positively contribute to some of the beneficial effects of the
FMD.
Note that 25% of the subjects who tested the FMD dropped out
of the trial, whereas 10% of the participants opted out of the control
arm. This indicates that, despite our efforts to reduce the burden of
low-calorie/protein diets, adherence to this dietary regimen requires
committed study participants. Further, compared to the control diet
arm, the FMD arm imposed an additional daylong visit to the clinic,
which may have contributed to reduced compliance. Compliance with
prescribed therapies, even placebo, may be an identifiable marker for
an overall healthy behavior of study participants (38). Thus, this kind
of volunteer, who is observing a benefit and thus not dropping out,
could introduce potential bias into the analysis of our trial. The overall
comparability at baseline between the control and both FMD arms, as
wellasthecomparableresponsetotheFMD(arm2andarm1after
crossover) suggests no major differences in response for those subjects
who completed the trial. Further, those subjects who dropped out of
this trial were not different in age or BMI compared to those who
completed the trial. It remains to be established why we experienced
a gender difference (82% of dropouts were female). The 25% overall
Fig. 3. Post hoc analysis of metabolic variables in subgroups identified by severity of risk factors. Subjects from both study arms who completed three FMD
cycles were post hoc stratified on the basis of being in either normal-risk or at-risk subgroups for factors associated with age-related diseases and conditions. The Dchange
shown represents comparisons to baseline. All data are means ± SD. Between-arm comparisons were calculated using two-tailed two-sample equal variance ttests. One-way
analysis of variance was used for the BMI groups (see Table 4 for details).
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Wei et al., Sci. Transl. Med. 9, eaai8700 (2017) 15 February 2017 9of12
by guest on October 20, 2017http://stm.sciencemag.org/Downloaded from
dropout rate (all causes) of study participants before the completion of
the third FMD cycle is in the range observed in other trials aimed at
evaluating dietary interventions in adults. For example, 16 weeks of
dieting in combination with physical exercise yielded a discontinuation
rate of about 30% (39), and a hypocaloric diet in 28 overweight/obese
womenresultedinadropoutrateof40%after6months(40). In a trial
assessing the effect of intermittent energy/carbohydrate restriction and
daily energy restriction on weight loss and metabolic disease risk markers
in overweight women, Harvie et al. reported a 23% dropout rate (41).
Nonetheless, there are limitations of our trial that should be considered:
(i) A relatively small number of subjects in the randomized comparison;
(ii) despite providing nourishment and calories for the duration of the
FMD, we experienced a higher dropout rate during the FMD interven-
tion than in the control arm; (iii) the findings that the FMD reduced
metabolic markers more effectively in at-risk subjects are based on a
non-randomized post hoc analysis of the individual factors in generally
healthy participants, and thus, further evaluation in subjects with diag-
nosed disease is needed.
Otherlessrestrictivedietssuchasthoserequiringaverylowcalorie
intake twice a week would impose 8 days per month of a severe restric-
tion compared to the 5 days per month or per several months of a less
restrictive intervention tested here (41). However, an advantage of these
diets is that they may not require as much medical supervision as the
longer FMD. FMDs or any type of prolonged fasting interventions last-
ing more than 12 hours, particularly those lasting several days, require
supervision, preferably from a health care professional familiar with pro-
longed fasting. Although our results suggestthatcyclesoftheplant-based
FMD might be safe for elderly individuals, additional studies are neces-
sary to determine its safety for subjects who are 70 years and older.
In summary, and with the limitations outlined above, these results
indicate that the periodic FMD cycles are effective in improving the
levels of an array of metabolic markers/risk factors associated with
poor health and aging and with multiple age-related diseases. As sug-
gested by preclinical studies, interventions that promote longevity
should also extend healthspan. Further investigations in larger clinical
trials focused on subjects with diagnosed metabolic syndrome, diabe-
tes, and CVDs as well as subjects at high risk for developing cancer
and other age-related diseases are needed.
MATERIALS AND METHODS
Subjects
One-hundred participants without a diagnosed medical condition in the
previous 6 months were enrolled (ClinicalTrials.gov; NCT02158897).
All participants provided written informed consent, and the University
of Southern California (USC) Institutional Review Board (IRB) ap-
proved the protocol. Recruitment of subjects was based on fliers, the
ClinicalTrials.gov and usc.com websites, and/or word of mouth. Be-
cause this was a dietary intervention study, it was not possible for par-
ticipants or all study personnel to be blinded to group assignment.
However, study personnel involved in data collection and specimen
analysis were blinded to group assignments.
Study design
Flow of participant enrollment and participation was prepared
following the CONSORT standards for randomized clinical trials
with crossover design. All data were collected at the USC Diabetes
and Obesity Research Institute. Subjects were recruited from April 2013
to July 2015 under protocols approved by the USC IRB (HS-12-00391)
based on established inclusion (generally healthy adult volunteers and
18 to 70 years of age; BMI, 18.5 and up) and exclusion [any major med-
ical condition or chronic diseases, mental illness, drug dependency, hor-
mone replacement therapy (dehydroepiandrosterone, estrogen, thyroid,
and testosterone), pregnant or nursing female, special dietary require-
ments or food allergies, alcohol dependency, and medications known to
affect body weight] criteria. Intention to treat analysis was performed by
including all available observations. Eligible participants were randomly
assigned using a random-number generator to either arm 1 or arm 2 of
the study. All participants completed a health habits questionnaire. Pre-
specified outcome measures included safety and feasibility, and evalua-
tion of changes in metabolic risk factors for diabetes and CVD and
metabolic markers associated with age-related diseases and mortality;
these outcomes were measured at baseline during and after completion
of the intervention. Laboratory examinations included height, weight,
body composition (including total and trunk body fat, soft lean tissue,
and bone mineral content) measured by dual-energy x-ray absorptiom-
etry (DEXA), oscillometric blood pressure measurements, and
overnight fasting blood draw through venipuncture.
Arm 1 (control).
Participants completed anthropometric measurements and blood collec-
tion at enrollment and after 3 months to provide an estimate of non–
diet-related changes (Fig. 1). Participants were instructed to maintain
their regular eating habits. After 3 months, subjects were crossed over
to the experimental FMD group (Fig. 1).
Arm 2 (FMD).
Participants were instructed to consume the FMD, which was provided
in a box, for 5 continuous days, and to return to their normal diet after
completion until the next cycle that was initiated about 25 days later.
Participants completed three cycles of this 5-day FMD (Fig. 1). Partic-
ipants completed baseline and follow-up examinations at the end of the
first FMD (before resuming normal diet to measure the acute FMD
effects) and after a washout period of 5 to 7 days of normal caloric
intake after the third FMD cycle. An optional follow-up assessment
3 months after the third FMD cycle was offered.
Experimental FMD
The FMD is a plant-based diet designed to attain fasting-like effects on
the serum levels of IGF-1, IGFBP-1, glucose, and ketone bodies while
providing both macro- and micronutrients to minimize the burden of
fasting and adverse effects (5). Day 1 of the FMD supplies ~4600 kJ
(11% protein, 46% fat, and 43% carbohydrate), whereas days 2 to 5 pro-
vide ~3000 kJ (9% protein, 44% fat, and 47% carbohydrate) per day.
The FMD comprises proprietary formulations belonging to USC and
L-Nutra (www.prolonfmd.com) of vegetable-based soups, energy bars,
energy drinks, chip snacks, tea, and a supplement providing high levels
of minerals, vitamins, and essential fattyacids(fig.S4).Allitemstobe
consumed per day were individually boxed to allow the subjects to
choose when to eat while avoiding accidentally consuming components
of the following day.
Common Terminology Criteria for Adverse Events
Study participants were asked about adverse events at each study visit;
events were graded according to the general CTCAE guidelines (see
the Supplementary Materials for details).
Blood tests and serum markers
Complete metabolic and lipid panels (overnight fasting) were completed
at the Clinical Laboratories at the Keck Medical Center of USC and
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Wei et al., Sci. Transl. Med. 9, eaai8700 (2017) 15 February 2017 10 of 12
by guest on October 20, 2017http://stm.sciencemag.org/Downloaded from
analyzed immediately after the blood draw of each visit (see the Sup-
plementary Materials for details).
Statistical analysis
The primary comparisons of randomized groups involved changes in
outcomes observed in the control period of arm 1 versus the changes
observed in the FMD group (arm 2) after completion of three FMD
cycles. Secondary observational analyses involved (i) comparing the
FMD effects in arm 2 (randomized to FMD) versus arm 1 (receiving
FMD after completion of the randomized control period) and (ii)
summarizing the changes for arms 1 and 2 combined after comple-
tion of the first and third FMD cycles. Changes from baseline were
normally distributed. Comparison of changes from baseline within
the treatment arms was performed using paired two-tailed Student’s
ttests, and Pvalues <0.05 were considered significant. The between-
arm comparison of treatment changes from baseline was performed
using two-tailed two-sample equal variance ttests, and Pvalues <0.05
were considered significant. To control for multiple testing, we used
the Benjamini-Hochberg FDR method. All reported Pvalues are
nominal two-sided Pvalues; those that met the FDR criteria and re-
mained “significant”at P< 0.05 are indicated with an asterisk.
M.W. generated the random allocation sequence and enrolled and
assigned participants to interventions. M.W. was not involved in out-
come assessments. For this initial randomized trial, the sample size of
100 total subjects was based on detection of a 25% reduction in mean
IGF-1, with a two-sided aof 0.05 and 70% power. The estimated con-
trol group mean (SD) IGF-1 of 194 (97) used published data on males
and females aged 26 to 40 years (42). Statistical analyses were per-
formed on deidentified data. Baseline information and changes from
baseline were summarized using means ± SDs for subjects randomized
to the control (arm 1, n=48)andthedietgroup(arm2,n=52).All
subjects are included in the arm assigned regardless of treatment adher-
ence (intention to treat); no attempt was made to impute missing values
(primarily because if data after completion of the third FMD cycle were not
available, then other measurement time points were usually unavailable).
In post hoc subgroup analyses, we compared FMD-control group
differences over the randomized trial period (three FMD cycles versus
control) within high/lower-risk subgroups and tested whether those
treatment effects differed in the higher-risk versus lower-risk groups.
This subgroup analysis was completed using analysis of variance, with
main effects of treatment (FMD and control) and risk group (high and
low); the interaction of treatment-by-risk group tested whether the
randomized FMD effect differed in high-risk versus low-risk groups.
In observational analyses of the pre-post FMD changes combining the
two treatment arms, pre-post changes in markers within risk subgroups
were tested using paired ttest; pre-post changes over risk subgroups
were compared using two-sample ttest or analysis of variance.
SUPPLEMENTARY MATERIALS
www.sciencetranslationalmedicine.org/cgi/content/full/9/377/eaai8700/DC1
Materials and Methods
Table S1. Complete metabolic panel.
Table S2. Arm-specific markers of adherence and changes in risk factors, including arm 1 after
crossover to FMD, and summary of FMD arms 1 and 2.
Table S3. Changes in risk factors and metabolic markers of adherence after the first FMD.
Table S4. Changesin risk factors and metabolic markers of adherence 3 months after intervention.
Fig. S1. Subject self-reported adverse effects based on CTCAE.
Fig. S2. Comparison of participants who completed the trial versus dropouts.
Fig. S3. Baseline to 3 months before/after comparison of individual subjects in the control
cohort and all subjects who completed the FMD.
Fig. S4. Nutritional information of the FMD.
CONSORT checklist
Trial protocols
REFERENCES AND NOTES
1. K.G.Alberti,R.H.Eckel,S.M.Grundy,P.Z.Zimmet,J.I.Cleeman,K.A.Donato,
J.-C. Fruchart, W. P. T. James, C. M. Loria, S. C. Smith Jr.; International Diabetes
Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and
Blood Institute; American Heart Association; World Heart Federation; International
Atherosclerosis Society; International Association for the Study of Obesity,
Harmonizing the metabolic syndrome: A joint interim statement of the International
Diabetes Federation Task Force on Epidemiology and Prevention; National Heart,
Lung, and Blood Institute; American Heart Association; World Heart Federation;
International Atherosclerosis Society; and International Association for the Study of
Obesity. Circulation 120,1640–1645 (2009).
2. E. S. Ford, W. H. Giles, W. H. Dietz, Prevalence of the metabolic syndrome among
US adults: Findings from the third National Health and Nutrition Examination Survey.
JAMA 287, 356–359 (2002).
3. A.S.Gami,B.J.Witt,D.E.Howard,P.J.Erwin,L.A.Gami,V.K.Somers,V.M.Montori,
Metabolic syndrome and risk of incident cardiovascular events and death: A
systematic review and meta-analysis of longitudinal studies. J. Am. Coll. Cardiol. 49,
403–414 (2007).
4. I. Y. Choi, L. Piccio, P. Childress, B. Bollman, A. Ghosh, S. Brandhorst, J. Suarez,
A. Michalsen, A. H. Cross, T. E. Morgan, M. Wei, F. Paul, M. Bock, V. D. Longo, A diet
mimicking fasting promotes regeneration and reduces autoimmunity and multiple
sclerosis symptoms. Cell Rep. 15, 2136–2146 (2016).
5. S. Brandhorst, I. Y. Choi, M. Wei, C. W. Cheng, S. Sedrakyan, G. Navarrete, L. Dubeau,
L. P. Yap, R. Park, M. Vinciguerra, S. D. Biase, H. Mirzaei, M. G. Mirisola, P. Childress, L. Ji,
S. Groshen, F. Penna, P. Odetti, L. Perin, P. S. Conti, Y. Ikeno, B. K. Kennedy, P. Cohen,
T. E. Morgan, T. B. Dorff, V. D. Longo, A periodic diet that mimics fasting promotes
multi-system regeneration, enhanced cognitive performance, and healthspan. Cell Metab.
22,86–99 (2015).
6. V. D. Longo, S. Panda, Fasting, circadian rhythms, and time-restricted feeding in healthy
lifespan. Cell Metab. 23, 1048–1059 (2016).
7. A. J. Bruce-Keller, G. Umberger, R. McFall, M. P. Mattson, Food restriction reduces
brain damage and improves behavioral outcome following excitotoxic and metabolic
insults. Ann. Neurol. 45,8–15 (1999).
8. A. L. Hartman, J. E. Rubenstein, E. H. Kossoff, Intermittent fasting: A “new”historical
strategy for controlling seizures? Epilepsy Res. 104, 275–279 (2013).
9. H. Müller, F. W. de Toledo, K. L. Resch, Fasting followed by vegetarian diet in patients with
rheumatoid arthritis: A systematic review. Scand. J. Rheumatol. 30,1–10 (2001).
10. V. D. Longo, M. P. Mattson, Fasting: Molecular mechanisms and clinical applications.
Cell Metab. 19, 181–192 (2014).
11. C.-W. Cheng, G. B. Adams, L. Perin, M. Wei, X. Zhou, B. S. Lam, S. D. Sacco, M. Mirisola,
D. I. Quinn, T. B. Dorff, J. J. Kopchick, V. D. Longo, Prolonged fasting reduces IGF-1/PKA
to promote hematopoietic-stem-cell-based regeneration and reverse immunosuppression.
Cell Stem Cell 14,810–823 (2014).
12. W.E.Sonntag,D.Lynch,W.T.Cefalu,R.L.Ingram,S.A.Bennett,P.L.Thornton,
A. S. Khan, Pleiotropic effects of growth hormone and insulin-like growth factor (IGF)-1 on
biological aging: Inferences from moderate caloric-restricted animals. J. Gerontol. A Biol.
Sci. Med. Sci. 54, B521–B538 (1999).
13. Y. Ikeno, R. T. Bronson, G. B. Hubbard, S. Lee, A. Bartke, Delayed occurrence of fatal
neoplastic diseases in Ames dwarf mice: Correlation to extended longevity. J. Gerontol.
A Biol. Sci. Med. Sci. 58, 291–296 (2003).
14. K. Flurkey, J. Papaconstantinou, R. A. Miller, D. E. Harrison, Lifespan extension and
delayed immune and collagen aging in mutant mice with defects in growth hormone
production. Proc. Natl. Acad. Sci. U.S.A. 98, 6736–6741 (2001).
15. S. E. Dunn, F. W. Kari, J. French, J. R. Leininger, G. Travlos, R. Wilson, J. C. Barrett, Dietary
restriction reduces insulin-like growth factor I levels, which modulates apoptosis, cell
proliferation, and tumor progression in p53-deficient mice. Cancer Res. 57, 4667–4672
(1997).
16. M. S. Bonkowski, F. P. Dominici, O. Arum, J. S. Rocha, K. A. Al Regaiey, R. Westbrook,
A. Spong, J. Panici, M. M. Masternak, J. J. Kopchick, A. Bartke, Disruption of growth
hormone receptor prevents calorie restriction from improving insulin action and
longevity. PLOS ONE 4, e4567 (2009).
17. M. E. Levine, J. A. Suarez, S. Brandhorst, P. Balasubramanian, C.-W. Cheng, F. Madia,
L. Fontana, M. G. Mirisola, J. Guevara-Aguirre, J. Wan, G. Passarino, B. K. Kennedy, M. Wei,
P. Cohen, E. M. Crimmins, V. D. Longo, Low protein intake is associated with a major
reduction in IGF-1, cancer, and overall mortality in the 65 and younger but not older
population. Cell Metab. 19, 407–417 (2014).
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Wei et al., Sci. Transl. Med. 9, eaai8700 (2017) 15 February 2017 11 of 12
by guest on October 20, 2017http://stm.sciencemag.org/Downloaded from
18. J. Guevara-Aguirre, P. Balasubramanian, M. Guevara-Aguirre, M. Wei, F. Madia,
C.-W. Cheng, D. Hwang, A. Martin-Montalvo, J. Saavedra, S. Ingles, R. de Cabo,
P. Cohe n, V. D. Longo, Growth hormone receptor deficiency is associated with a major
reduction in pro-aging signaling, cancer, and diabetes in humans. Sci. Transl. Med. 3,
70ra13 (2011).
19. www.census.gov
20. M. W. Gillman, Developmental origins of health and disease. N. Engl. J. Med. 353,
1848–1850 (2005).
21. L. Fontana, B. K. Kennedy, V. D. Longo, D. Seals, S. Melov, Medical research: Treat ageing.
Nature 511, 405–407 (2014).
22. M. Nayor, R. S. Vasan, Recent update to the US cholesterol treatment guidelines:
A comparison with international guidelines. Circulation 133, 1795–1806 (2016).
23. American Diabetes Association, Diagnosis and classification of diabetes mellitus.
Diabetes Care 37 (suppl. 1), S81–S90 (2014).
24. M. Miller, N. J. Stone, C. Ballantyne, V. Bittner, M. H. Criqui, H. N. Ginsberg, A. C. Goldberg,
W. J. Howard, M. S. Jacobson, P. M. Kris-Etherton, T. A. Lennie, M. Levi, T. Mazzone,
S. Pennathur, Triglycerides and cardiovascular disease: A scientific statement from the
American Heart Association. Circulation 123, 2292–2333 (2011).
25. T. A. Pearson, G. A. Mensah, R. W. Alexander, J. L. Anderson, R. O. Cannon, M. Criqui,
Y. Y. Fadl, S. P. Fortmann, Y. Hong, G. L. Myers, N. Rifai, S. C. Smith, K. Taubert, R. P. Tracy,
F. Vinicor; Centers for Disease Control and Prevention, American Heart Association,
Markers of inflammation and cardiovascular disease: Application to clinical and
public health practice: A statement for healthcare professionals from the Centers
for Disease Control and Prevention and the American Heart Association. Circulation
107,499–511 (2003).
26. M. N. Pollack, Insulin, insulin-like growth factors, insulin resistance, and neoplasia. Am. J.
Clin. Nutr. 86, s820–s822 (2007).
27. M. E. Levine, Modeling the rate of senescence: Can estimated biological age predict
mortality more accurately than chronological age? J. Gerontol. A Biol. Sci. Med. Sci. 68,
667–674 (2013).
28. L. Fontana, L. Partridge, V. D. Longo, Extending healthy life span—From yeast to humans.
Science 328, 321–326 (2010).
29. S. Brandhorst, M. Wei, S. Hwang, T. E. Morgan, V. D. Longo, Short-term calorie and protein
restriction provide partial protection from chemotoxicity but do not delay glioma
progression. Exp. Gerontol. 48, 1120–1128 (2013).
30. A. G. Renehan, M. Zwahlen, C. Minder, S. T. O’Dwyer, S. M. Shalet, M. Egger, Insulin-like
growth factor (IGF)-I, IGF binding protein-3, and cancer risk: Systematic review and
meta-regression analysis. Lancet 363, 1346–1353 (2004).
31. J. M. Chan, M. J. Stampfer, J. Ma, P. Gann, J. M. Gaziano, M. Pollak, E. Giovannucci,
Insulin-like growth factor-I (IGF-I) and IGF binding protein-3 as predictors of
advanced-stage prostate cancer. J. Natl. Cancer Inst. 94, 1099–1106 (2002).
32. N.E.Allen,A.W.Roddam,D.S.Allen,I.S.Fentiman,I.dosSantosSilva,J.Peto,
J. M. P. Holly, T. J. Key, A prospective study of serum insulin-like growth factor-I
(IGF-I), IGF-II, IGF-binding protein-3 and breast cancer risk. Br. J. Cancer 92,
1283–1287 (2005).
33. O. Fletcher, L. Gibson, N. Johnson, D. R. Altmann, J. M. P. Holly, A. Ashworth, J. Peto,
I. dos Santos Silva, Polymorphisms and circulating levels in the insulin-like growth factor
system and risk of breast cancer: A systematic review. Cancer Epidemiol. Biomarkers Prev.
14,2–19 (2005).
34. K.Rapp,J.Schroeder,J.Klenk,H.Ulmer,H.Concin,G.Diem,W.Oberaigner,
S. K. Weiland, Fasting blood glucose and cancer risk in a cohort of more than 140,000
adults in Austria. Diabetologia 49, 945–952 (2006).
35. T. Stocks, K. Rapp, T. Bjørge, J. Manjer, H. Ulmer, R. Selmer, A. Lukanova, D. Johansen,
H. Concin, S. Tretli, G. Hallmans, H. Jonsson, P. Stattin, Blood glucose and risk of incident
and fatal cancer in the metabolic syndrome and cancer project (Me-Can): Analysis of
six prospective cohorts. PLOS Med. 6, e1000201 (2009).
36. M. N. Harvie, M. Pegington, M. P. Mattson, J. Frystyk, B. Dillon, G. Evans, J. Cuzick,
S. A. Jebb, B. Martin, R. G. Cutler, T. G. Son, S. Maudsley, O. D. Carlson, J. M. Egan,
A. Flyvbjerg, A. Howell, The effects of intermittent or continuous energy restriction on
weight loss and metabolic disease risk markers: A randomized trial in young overweight
women. Int. J. Obes. 35, 714–727 (2011).
37. L. Fontana, E. P. Weiss, D. T. Villareal, S. Klein, J. O. Holloszy, Long-term effects of calorie or
protein restriction on serum IGF-1 and IGFBP-3 concentration in humans. Aging Cell 7,
681–687 (2008).
38. R. D. Hays, R. L. Kravitz, R. M. Mazel, C. D. Sherbourne, M. R. DiMatteo, W. H. Rogers,
S. Greenfield, The impact of patient adherence on health outcomes for patients with
chronic disease in the medical outcomes study. J. Behav. Med. 17, 347–360 (1994).
39. T. P. Wycherley, M. Noakes, P. M. Clifton, X. Cleanthous, J. B. Keogh, G. D. Brinkworth,
A high-protein diet with resistance exercise training improves weight loss and body
composition in overweight and obese patients with type 2 diabetes. Diabetes Care 33,
969–976 (2010).
40. D. Florakis, E. Diamanti-Kandarakis, I. Katsikis, G. P. Nassis, A. Karkanaki, N. Georgopoulos,
D. Panidis, Effect of hypocaloric diet plus sibutramine treatment on hormonal and
metabolic features in overweight and obese women with polycystic ovary syndrome:
A randomized, 24-week study. Int. J. Obes. 32, 692–699 (2008).
41. M. Harvie, C. Wright, M. Pegington, D. McMullan, E. Mitchell, B. Martin, R. G. Cutler,
G. Evans, S. Whiteside, S. Maudsley, S. Camandola, R. Wang, O. D. Carlson, J. M. Egan,
M. P. Mattson, A. Howell, The effect of intermittent energy and carbohydrate restriction v.
daily energy restriction on weight loss and metabolic disease risk markers in overweight
women. Br. J. Nutr. 110, 1534–1547 (2013).
42. G. Brabant, A. von zur Mühlen, C. Wüster, M. B. Ranke, J. Kratzsch, W. Kiess, J.-M. Ketelslegers,
L. Wilhelmsen, L. Hulthén, B. Saller, A. Mattsson, J. Wilde, R. Schemer, P. Kann, Serum
insulin-like growth factor I reference values for an automated chemiluminescence
immunoassay system: Results from a multicenter study. Horm. Res. 60,53–60 (2003).
Acknowledgments: We thank Y. Guan for help in data curation and acknowledge the
technical expertise of R. Buono (USC Leonard Davis School of Gerontology). Funding: Funding
was provided by the USC Edna Jones chair fund to V.D.L. W.J.M.’s contributions were provided
through the Southern California Clinical and Translational Science Institute supported by
NIH UL1TR001855. Author contributions: V.D.L., M.W., T.E.M., and T.D. designed the clinical
trial. M.W., K.H., and T.D. supervised the clinical trial. J.B. and M.S. contributed to patient
recruitment and supervision. S.B., H.M., C.W.C., E.G., S.G., and W.J.M. performed data analysis.
S.B. and V.D.L. wrote the paper. All authors discussed the r esults and commente d on the
manuscript. Co mpeting intere sts: The experimen tal FMD was provided by L-Nutra Inc. The
funding sources had no involvement in study des ign; collection, analysis, and interpretation
of data; writing of t he report; or decisi on to submit the article for publicati on. The USC has
licensed intell ectual property to L-Nutra that is under stu dy in this research.
As part of this license agreeme nt, the University has the potential to receive royalty
payments from L-Nutra. V.D.L. a nd T.E.M., who have equity interes t in L-Nutra, did no t
participate in the collection and analysis of the data. One-hundred percent of V.D.L.’s equity
will be assigne d to the nonprofit foundation Create Cures. U.S. pa tents 20140227 373A1
(“Methods and diets to protec t against chemot oxic ity and age related illnesses”)and
20140112909A1 (“Methods and formulations promoting tissue regeneration, longevity and
healthspan”) related to the work described here have been filed. Data and materialsavailability:
Use of the FMD may require a material transfer agreement.
Submitted 27 April 2016
Resubmitted 23 August 2016
Accepted 20 December 2016
Published 15 February 2017
10.1126/scitranslmed.aai8700
Citation: M. Wei, S. Brandhorst, M. Shelehchi, H. Mirzaei, C. W. Cheng, J. Budniak, S. Groshen,
W. J. Mack, E. Guen, S. Di Biase, P. Cohen, T. E. Morgan, T. Dorff, K. Hong, A. Michalsen,
A. Laviano, V. D. Longo, Fasting-mimicking diet and markers/risk factors for aging, diabetes,
cancer, and cardiovascular disease. Sci. Transl. Med. 9, eaai8700 (2017).
SCIENCE TRANSLATIONAL MEDICINE |RESEARCH ARTICLE
Wei et al., Sci. Transl. Med. 9, eaai8700 (2017) 15 February 2017 12 of 12
by guest on October 20, 2017http://stm.sciencemag.org/Downloaded from
cardiovascular disease
Fasting-mimicking diet and markers/risk factors for aging, diabetes, cancer, and
Michalsen, Alessandro Laviano and Valter D. Longo
Wendy J. Mack, Esra Guen, Stefano Di Biase, Pinchas Cohen, Todd E. Morgan, Tanya Dorff, Kurt Hong, Andreas
Min Wei, Sebastian Brandhorst, Mahshid Shelehchi, Hamed Mirzaei, Chia Wei Cheng, Julia Budniak, Susan Groshen,
DOI: 10.1126/scitranslmed.aai8700
, eaai8700.9Sci Transl Med
healthy metabolic system.
study is needed to replicate these results, but they raise the possibility that fasting may be a practical road to a
effects were generally larger in the subjects who were at greater risk of disease at the start of the study. A larger
decreased BMI, glucose, triglycerides, cholesterol, and C-reactive protein (a marker for inflammation). These
been implicated in aging and disease. A post hoc analysis replicated these results and also showed that fasting
diet reduced body weight and body fat, lowered blood pressure, and decreased the hormone IGF-1, which has
months or maintained their normal diet for 3 months and then switched to the fasting schedule. The fasting-like
studied 71 people who either consumed a fasting-mimicking diet for 5 days each month for 3et al.as well, Wei
Mice that fast periodically are healthier, metabolically speaking. To explore whether fasting can help people
Fasting: More than a fad
ARTICLE TOOLS http://stm.sciencemag.org/content/9/377/eaai8700
MATERIALS
SUPPLEMENTARY http://stm.sciencemag.org/content/suppl/2017/02/13/9.377.eaai8700.DC1
CONTENT
RELATED
http://stm.sciencemag.org/content/scitransmed/9/407/eaad4000.full
http://science.sciencemag.org/content/sci/357/6350/455.full
http://science.sciencemag.org/content/sci/357/6350/507.full
http://stm.sciencemag.org/content/scitransmed/9/394/eaah4477.full
REFERENCES http://stm.sciencemag.org/content/9/377/eaai8700#BIBL
This article cites 41 articles, 12 of which you can access for free
PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions
Terms of ServiceUse of this article is subject to the
is a registered trademark of AAAS.Science Translational Medicinetitle
licensee American Association for the Advancement of Science. No claim to original U.S. Government Works. The
Science, 1200 New York Avenue NW, Washington, DC 20005. 2017 © The Authors, some rights reserved; exclusive
(ISSN 1946-6242) is published by the American Association for the Advancement ofScience Translational Medicine
by guest on October 20, 2017http://stm.sciencemag.org/Downloaded from
