Science - USA (2021-12-24)

of fiber on antitumor immunity by affecting
pathways of T cell activation as well as the
accumulation of T cells in the tumor, including
inducible T cell co-stimulator (ICOS)–expressing
CD8+and CD4+T cells (fig. S17, C to H).
Together, these data have important impli-
cations. We show that dietary fiber and pro-
biotic use, factors known to affect the gut
microbiome, are associated with differential
outcomes to ICB. Although causality cannot
be addressed from the observational human
cohort, where unmeasured confounders may
exist, our preclinical models support the hy-
pothesis that dietary fiber and probiotics mod-
ulate the microbiome and that antitumor
immunity is impaired in mice receiving a low-
fiber diet and in those receiving probiotics—
with suppression of intratumoral IFN-gT cell
responses in both cases.
Numerous challenges exist to decipher how
best to leverage the microbiome to optimize
patient outcomes, starting with what to target—
selected features or community function—and
whether this can be safely achieved through
supplementation or more comprehensive die-
tary approaches. Several prior studies have
shown that controlled increases in dietary fiber
intake can modulate the gut microbiome but
also that interindividual variation in the gut
microbiome drives differential effects of spe-
cific fibers (and prebiotics) on host metabolism
( 36 – 40 ). Ongoing dietary intervention studies
in the setting of ICB are critical for establishing
whether a targeted and achievable diet change
at the initiation of ICB can safely and effectively
improve outcomes (NCT04645680). Although
our findings suggest that undirected use of
commercially available probiotics may be harm-
ful in the setting of ICB, further study of
rationally designed and targeted probiotics
or bacterial consortia is warranted on the
basis of promising early data of this approach
( 22 – 24 ).
Some analyses in the current cohort were
not adequately powered to assess the full ef-
fect of these factors, and further validation is
needed in independent cohorts with more in-
depth and detailed assessment of dietary intake

and the use of specific probiotic supplements,

along with further mechanistic studies in pre-

clinical models. Nonetheless, these notable (and

perhaps unexpected) findings from studies in

this observational patient cohort are corrobo-

rated by parallel studies in preclinical models

with preliminary mechanistic insights. In light

of these collective results, dietary habits and

probiotic supplement use should be consi-

dered in patients receiving ICB and in efforts

to modulate the gut microbiota. These factors

should be more thoughtfully evaluated in

strategies to improve cancer outcomes.

REFERENCESANDNOTES

1. A. Ribas, J. D. Wolchok,Science 359 , 1350–1355 (2018).


2. B. Routyet al.,Science 359 , 91–97 (2018).


3. V. Matsonet al.,Science 359 , 104–108 (2018).


4. V. Gopalakrishnanet al.,Science 359 , 97–103 (2018).


5. A. E. Frankelet al.,Neoplasia 19 , 848–855 (2017).


6. B. A. Peterset al.,Genome Med. 11 , 61 (2019).


7. M. C. Andrewset al.,Nat. Med. 27 , 1432–1441 (2021).


8. G. D. Wuet al.,Science 334 , 105–108 (2011).


9. D. McDonaldet al.,mSystems 3 , e00031-18 (2018).


10. L. Maieret al.,Nature 555 , 623–628 (2018).


11. J. Suezet al.,Cell 174 , 1406–1423.e16 (2018).


12. B. P. Willing, S. L. Russell, B. B. Finlay,Nat. Rev. Microbiol. 9 ,
    233 – 243 (2011).


13. F. Asnicaret al.,Nat. Med. 27 , 321–332 (2021).


14. D. Rothschildet al.,Nature 555 , 210–215 (2018).


15. E. A. Eisenhaueret al.,Eur. J. Cancer 45 , 228– 247
    (2009).


16. F. E. Thompson, D. Midthune, L. Kahle, K. W. Dodd,J. Nutr.
    147 , 1226–1233 (2017).


17. National Cancer Institute, Dietary Screener Questionnaire
    (DSQ) in the NHANES 2009-10: Data processing and scoring
    procedures (2019); https://epi.grants.cancer.gov/nhanes/
    dietscreen/scoring.


18. R. Z. Gharaibeh, C. Jobin,Gut 68 , 385–388 (2019).


19. E. N. Baruchet al.,Science 371 , 602–609 (2021).


20. D. Davaret al.,Science 371 , 595–602 (2021).


21. J. C. Arthuret al.,Sci. Rep. 3 , 2868 (2013).


22. Y. Tomitaet al.,Cancer Immunol. Res. 8 , 1236– 1242
    (2020).


23. A. Sivanet al.,Science 350 , 1084–1089 (2015).


24. T. Tanoueet al.,Nature 565 , 600–605 (2019).


25. H. J. Flint, K. P. Scott, S. H. Duncan, P. Louis, E. Forano,
    Gut Microbes 3 , 289–306 (2012).


26. G. den Bestenet al.,J. Lipid Res. 54 , 2325– 2340
    (2013).


27. S. A. Poekeret al.,Sci. Rep. 8 , 4318 (2018).


28. D. Soet al.,Am. J. Clin. Nutr. 107 , 965–983 (2018).


29. M. S. Desaiet al.,Cell 167 , 1339–1353.e21 (2016).


30. J. L. McQuadeet al.,Lancet Oncol. 19 , 310–322 (2018).


31. Y. Anet al.,J. Transl. Med. 18 , 235 (2020).


32. D. J. Pinatoet al.,JAMA Oncol. 5 , 1774–1778 (2019).


33. D. M. Klurfeldet al.,Am. J. Physiol. Endocrinol. Metab. 315 ,
    E1087–E1097 (2018).
    34. K. Makki, E. C. Deehan, J. Walter, F. Bäckhed,Cell Host Microbe
    23 , 705–715 (2018).
    35. E. Pérez-Guijarroet al.,Nat. Med. 26 , 781–791 (2020).
    36. L. Michalaket al.,Nat. Commun. 11 , 5773 (2020).
    37. A. Oliveret al.,mSystems 6 , e00115-21 (2021).
    38. A. Leshem, E. Segal, E. Elinav,mSystems 5 , e00665-20 (2020).
    39. S. M. Murga-Garridoet al.,Microbiome 9 , 117 (2021).
    40. C. R. Daniel, J. L. McQuade,Trends Cancer 5 , 521–524 (2019).

ACKNOWLEDGMENTS

C.N.S. acknowledges the Parker Institute for Cancer

Immunotherapy for funding time devoted to continued analysis

of this project. She also acknowledges M. D. Swartz, L. B. Piller, and

X. Du at UTSPH for their participation on her thesis committee.

J.L.M. acknowledges the Transdisciplinary Research in Energetics

and Cancer Research Training Workshop R25CA203650. J.L.M. and

C.R.D. acknowledge the MDACC Center for Energy Balance in

Cancer Prevention and Survivorship. C.R.D. acknowledges the

MDACC Bionutrition Core. We thank our collaborators and

personnel at the LICI, NCI Microbiome and Genetics Core Facility,

and NCI Mouse Gnotobiotic Core Facility, The Alkek Center for

Metagenomics and Microbiome Research (CMMR) at Baylor

College of Medicine (including S. J. Javornik Cregeen for her role in

microbiome data processing), and CosmosID for their high

quality and timely microbiome data generation services. We also

wish to acknowledge MD Anderson’s Program for Innovative

Microbiome and Translational Research (PRIME-TR) for supporting

the analysis and interpretation of the microbiome results

presented herein (J.A.W. and N.J.A. are the program director and

executive scientific director for PRIME-TR, respectively). Most

importantly, the study team wishes to thank all patients who

contributed their time, samples, and data to this research.

Funding:This study received support from National Institute of

Health grant 1R01 CA219896-01A1 (J.A.W.); US-Israel Binational

Science Foundation grant 201332 (J.A.W.); the Melanoma

Research Alliance grant 4022024 (J.A.W.); American Association

for Cancer Research Stand Up to Cancer grant SU2C-AACR-

IRG-19-17 (J.A.W.); the Andrew Sabin Family Fellows Program

(J.A.W. and C.R.D.); MD Anderson Cancer Center’s Melanoma

Moon Shots Program (J.A.W., J.L.M., C.R.D., M.A.D., H.A.T., J.E.G.,

and E.M.B.); and the Melanoma Research Alliance grant 564449

(L.C., J.A.W., and J.L.M.). The authors additionally received support

from Department of Defense grant W81XWH 16 1 0121 (J.A.W.); the

MD Anderson Cancer Center Multidisciplinary Research Program

grant (J.A.W.); the Parker Institute for Cancer Immunotherapy at

MD Anderson Cancer Center (J.A.W., H.A.T., P.S., and J.P.A.);

American Society of Clinical Oncology and Conquer Cancer

Foundation Career Development award AWD0000627 (J.L.M.); the

Elkins Foundation (J.L.M.); the Seerave Foundation (J.L.M.);

Rising Tide Foundation grant AWD00004505 (J.L.M.); the Mark

Foundation grant AWD00004538 (J.L.M.); the Longenbaugh-Torian

Fund (J.L.M.); the MD Anderson Cancer Center SPORE in

Melanoma P50CA221703 (M.A.D., H.A.T., I.C.G., J.A.W., S.P.P.,

J.E.L., J.E.G., A.J.L., and J.L.M.); the MD Anderson Physician

Scientist Program (J.L.M.); the Cancer Research Institute

Irvington Fellowship Program (M.V.); Cancer Prevention and

Research Institute of Texas Research Training Program RP170067

(A.P.C., J.T., and J.Z.); Cancer Prevention and Research Institute

of Texas Research Training Program RP210028 (L.M.K.); the

US Department of State, Bureau of Educational and Cultural Affairs

(A.P.C.); the Fulbright Franco–Américaine Commission (A.P.C.);

Cancer Prevention and Research Institute of Texas Research

SCIENCEscience.org 24 DECEMBER 2021•VOL 374 ISSUE 6575 1639

Fig. 3. Effect of dietary fiber intake in patients and in preclinical models
of melanoma immunotherapy.(A) Kaplan-Meier plot comparing progression-
free survival intervals by dietary fiber intake among patients who received
ICB (n= 128;P= 0.047 by log-rank test). (B) Kaplan-Meier plot comparing
progression-free survival intervals by combined dietary fiber and probiotic
status among patients who received ICB (n= 123; overallPacross four
groups = 0.11;Pfor sufficient dietary fiber intake + no probiotics use versus
else = 0.015; both by log-rank test). (C) Experimental design of studies in
C57BL/6 SPF mice that received either a high-fiber or low-fiber diet at
inoculation of M3 (HCmel1274) melanoma cells (1 × 10^6 tumor cells) and
were then treated with anti–PD-1 or isotype control. Time is in days relative
to tumor injection. (D) M3 melanoma growth kinetics of control (high-fiber)
diet (circles) or low-fiber (fiber-free) diet (squares) treated four times with
intraperitoneal injection of anti–PD-1 antibody (dark green) or of isotype

control (Iso Ctrl) (light green). Data are means ± SEM of tumor volume from

one representative experiment (n=5pergroup).AllPvalues are from a

likelihood ratio test in a linear mixed model (isotype control and high fiber,

P= 0.69; anti–PD-1 and high fiber,P= 0.02; anti–PD-1 and low fiber versus

isotype control and low fiber,P= 0.08). *P< 0.05. (E)t-UMAPplot

comparing the gut microbiome (via shotgun metagenomic sequencing of

fecal samples) of mice by treatment and diet group from two experiments

(n= 4 to 5 per group) using Bray-Curtis distances (PERMANOVAP< 0.0001)

at experimental day 16. (F) Heatmap of gene expression of flow-sorted

CD45+tumor-infiltrating immunocytes in mice fed high- versus low-fiber diets

and treated with anti–PD-1 or isotype control. (G) Gene set enrichment

analysis depicting pathways enriched in high-fiber diet mice treated with

anti–PD-1 versus isotype control which were not differentially expressed by

treatment in low-fiber diet mice.

RESEARCH | REPORTS