1
A C i ical Commen a y on “An i-p oges in he apy
a ge s hallma ks o b eas cance isk” by Simoes
e al. Na u e 2025; doi:10.1038/s41586-025-09684-
7
Mengxi Zhu and Shu-Feng Zhou*
College o Chemical Enginee ing, Huaqiao Uni e si y, Xiamen, China
Co espondence: [email p o ec ed]
In oduc ion
Simoes e al.1 p esen a high-p o ile Na u e s udy p oposing an i-p oges in he apy as a
p e en i e s a egy o b eas cance , g ounded in he concep ha p oges e one
signaling con ibu es o ea ly umo -ini ia ing e en s in he human b eas . The au ho s
combine ex i o b eas issue assays, gene ically enginee ed mouse models, pa ien -
de i ed samples, and sys ems-le el analysis o claim ha an i-p oges ins supp ess
se e al “hallma ks o b eas cance isk”—including epi helial p oli e a ion, s em-like
cell expansion, mic oen i onmen al p iming, and p oges e one- esponsi e
ansc ip ional p og ams.
While he opic is imely and clinically impo an , he s udy aises me hodological,
in e p e a i e, and ep oducibili y conce ns, especially in he igu e da a. Se e al igu es
appea unde -con olled, s a is ically agile, o o e -in e p e ed, and some Ex ended
Da a/Supplemen a y Figu es aise image-quali y o quan i ica ion anspa ency issues
ha esemble o he PubPee - lagged pa e ns in mamma y gland biology s udies.
Below, we o e a igu e-by- igu e c i ique, ollowed by in-dep h e alua ions o Ex ended
Da a and Supplemen a y Figu es.
Figu e-by-Figu e C i ique
Figu e 1 – Baseline a iabili y and insu icien con ols
unde mine he cen al claim
Figu e 1 a emp s o show ha an i-p oges in ea men educes p oli e a i e and s em-
like compa men s in human b eas issue explan s. Howe e , he expe imen al design
displays se e al laws:
2
1.1. Dono he e ogenei y no adequa ely con olled
Human b eas explan s inhe en ly exhibi ex eme dono - o-dono a iabili y in pa i y,
BMI, mens ual cycle s age, gene ic backg ound, ho monal milieu, and issue
composi ion. The au ho s appea o use n = 4–6 dono s, which is a guably
unde powe ed o claim obus gene alizable e ec s.
The sca e plo s show wide in a-dono a iabili y, ye s a is ical compa isons ea all
da a poin s as independen biological eplica es. Wi hou pai ed dono -wise analysis,
conclusions lack igo .
1.2. Immuno luo escence panels show inconsis encies
The p oli e a ion ma ke Ki-67 s aining exhibi s:
• Une en illumina ion
• Va iable backg ound luo escence
• Di e en exposu e le els be ween con ol and ea men
These inconsis encies comp omise quan i a i e in e p e a ion. The manusc ip does no
p o ide he aw image exposu e se ings o de ails ega ding ba ch no maliza ion.
1.3. Ques ionable use o su oga e s emness ma ke s
The au ho s ely hea ily on ALDH ac i i y, a no o iously non-speci ic ma ke in luenced
by me abolic s a e and apop osis. An i-p oges ins may al e me abolism independen o
s emness, con ounding in e p e a ion.
1.4. Po en ial pseudo eplica ion
Mul iple o ganoids om a single dono a e ea ed as independen eplica es (“n”
coun ing p oblem), a i icially in la ing s a is ical signi icance.
Conclusion o Figu e 1: The da a may be di ec ionally in e es ing, bu he s a is ical
analysis is weak and he imaging/quan i ica ion app oaches a e no su icien ly
s anda dized o suppo s ong claims.
Figu e 2 – Mouse model da a o e s a ed, wi h poo alignmen o
human biology
Figu e 2 p esen s mouse model da a claiming ha an i-p oges in he apy educes ea ly
p e-neoplas ic ea u es.
3
2.1. Inadequa e jus i ica ion o he mouse model
The s udy uses p oges e one-d i en mamma y ou g ow h models, ye he dose and
schedule o ho mone adminis a ion do no e lec endogenous ho monal cycling in
humans.
The au ho s ail o discuss:
• Di e ences in p oges e one ecep o iso o m exp ession
• Species-speci ic s omal–epi helial in e ac ions
• The unna u al ho mone exposu e egimen
2.2. Imaging quali y conce ns
Some whole-moun mamma y gland images exhibi :
• O e -con as adjus men s
• Po en ial sa u a ion o epi helial duc s
• Suspiciously simila b anching pa e ns be ween samples (possible euse o
images o o e -alignmen )
2.3. Quan i ica ion de i ed om e y small egions
The duc al b anching and side-b anching analyses appea o use iny c opped egions,
igno ing gland-le el he e ogenei y. Wi hou gland-wide quan i ica ion, he conclusions
a e enuous.
2.4. Tumo p e en ion claims a e p ema u e
While he au ho s show educ ions in ea ly p eneoplas ic lesions, hey do no p esen
longi udinal umo incidence cu es o Kaplan–Meie analysis. T ansla ing educ ion in
“ isk hallma ks” o ue cance p e en ion is specula i e.
Figu e 3 – T ansc ip ional p o iling aises ep oducibili y and
ba ch-e ec conce ns
The RNA-seq signa u e analysis is a majo mechanis ic pilla o he s udy. Un o una ely,
he igu e and me hods aise mul iple conce ns.
3.1. P incipal componen analysis (PCA) sugges s ba ch e ec s
The PCA plo shows samples clus e ing by ba ch a he han ea men . The au ho s
a ibu e his o “dono -le el di e ences”, bu close inspec ion sugges s po en ial
lib a y p epa a ion o sequencing ba ch con ounding.
4
3.2. O e -in e p e a ion o gene se en ichmen
The au ho s claim ha an i-p oges ins supp ess “b eas cance isk hallma ks”. Howe e ,
he GSEA e ms appea o be b oad s ess- esponse pa hways, many o which can be
induced by any cy os a ic d ug.
Wi hou speci ic p oges e one- a ge gene con ols, he mechanis ic speci ici y is no
demons a ed.
3.3. Lack o alida ion in independen da ase s
The au ho s do no alida e he ansc ip omic signa u es agains publicly a ailable
da ase s such as:
• GTEx mamma y issue
• Single-cell a lases
• P oges e one- ea ed p ima y o ganoids
This is a majo missed oppo uni y and unde mines he uni e sali y o he claims.
Figu e 4 – Mic oen i onmen and immune signaling da a lack
quan i ica ion anspa ency
Figu e 4 claims ha an i-p oges ins modula e ib oblas ac i a ion, ex acellula ma ix
emodeling, and immune-cell ec ui men .
Howe e :
4.1. Immuno luo escence lacks consis en acquisi ion pa ame e s
Fib oblas ma ke s (α-SMA, FAP) show subs an ial di e ences in:
• Exposu e
• Sha pness
• Backg ound le els
• Field o iew
No aw images o imaging me ada a a e p o ided.
4.2. Suspec ed duplica ed egions
In wo panels, s omal egions appea suspiciously epe i i e—possibly he same egion
imaged unde di e en channels. This equi es cla i ica ion.
4.3. Ambiguous immune in il a ion quan i ica ion
The quan i ica ion ba s lack cla i y on:
5
• Whe he coun s a e pe ield, pe mm², o pe gland
• Whe he images we e andomized
• Whe he image analysis was blinded
Gi en he p opensi y o s omal ea u es o clus e , wi hou igo ous sampling, he
immune-cell-densi y da a could be hea ily biased.
4.4. Con ols missing o sys emic ho monal e ec s
An i-p oges ins ha e sys emic immunomodula o y e ec s, ye he au ho s p esen he
immune-cell changes as mamma y-speci ic, wi hou measu ing:
• Ci cula ing cy okines
• Spleen o lymph node immune popula ions
• O - a ge hema ologic impac s o ea men
Figu e 5 – T ansla ional claims exceed he da a
Figu e 5 is he caps one, p esen ing he s udy’s clinical implica ions.
5.1. O e -ex apola ion om sho - e m bioma ke changes
The au ho s claim an i-p oges ins can “ educe b eas cance isk”, bu he da a show
only:
• Reduced epi helial p oli e a ion
• Al e ed s omal ma ke s
• T ansc ip ional modula ion
These a e su oga e ma ke s, no alida ed p edic o s o ac ual cance incidence.
5.2. Lack o s a is ical co ec ion o mul iple compa isons
The igu e compiles dozens o indi idual ma ke s ac oss se e al expe imen al sys ems,
ye co ec ions o mul iple es ing a e no clea ly s a ed.
5.3. Missing pha macokine ic/pha macodynamic da a
Fo a ansla ional “ he apy”- amed claim, no PK/PD cu es o a ge -engagemen
assays a e p esen ed.
5.4. Po en ial con lic s wi h exis ing clinical ial da a
P io ials o an i-p oges ins o b eas cance p e en ion (e.g., mi ep is one) ha e
shown limi ed e icacy. The au ho s do no econcile hei model wi h he b oade
li e a u e.
6
Ex ended Da a Figu es – C i ical E alua ion
Ex ended Da a Figu e 1 – Insu icien dono me ada a
The au ho s p o ide a dono -le el able bu omi key a iables:
• Mens ual cycle phase
• Se um p oges e one/es ogen le els
• BRCA s a us
• O al con acep i e use
• Deg ee o adiposi y in he sampled issue
These a iables hea ily in luence p oges e one esponsi eness. The absence o his
in o ma ion se e ely weakens he in e p e a ion o human explan esponses.
Ex ended Da a Figu e 2 – Ques ionable no maliza ion me hods
The ALDH low cy ome y ex ended da a appea o be no malized o “% o con ol”, bu
he au ho s do no disclose:
• Raw luo escence in ensi ies
• Ga ing s a egies
• Compensa ion con ols
The lack o a p ope ga ing ee makes i impossible o e alua e whe he he ALDH-high
popula ion is accu a ely sepa a ed om deb is, au o luo escence, o apop o ic cells.
Ex ended Da a Figu e 3 – Con lic ing p oli e a ion quan i ica ion
me hods
The igu e compa es:
• Ki-67 immunos aining
• EdU inco po a ion
• Cyclin D1 posi i i y
Bu he h ee assays show inconsis encies, sugges ing ei he :
• Technical noise
• Di e ences in sample handling
• Asynch onous p oli e a i e esponses
The au ho s selec i ely emphasize he assays a o able o hei hypo hesis.
7
Ex ended Da a Figu e 4 – O e -p ocessed whole-moun images
The mamma y gland whole-moun s appea hea ily p ocessed:
• Edges appea compu a ionally sha pened
• Gland bounda ies look a i icially smoo h
• Pixel noise pa e ns sugges pos -hoc il e ing
Such image manipula ion, in en ional o no , complica es biological in e p e a ion.
Ex ended Da a Figu e 5 – Bulk RNA-seq QC missing
The au ho s do no p o ide:
• Read-dep h s a is ics
• Pe cen age alignmen
• Duplica e ead pe cen ages
• Mi ochond ial o ibosomal ead con en
In high-p o ile genomics pape s, hese a e s anda d and should be manda o y. The
absence aises ques ions abou da ase quali y.
Ex ended Da a Figu e 6 – GSEA olcano plo conce ns
The olcano plo s appea :
• O e -symme ic (a ed lag 🚩 o ba ch-d i en DE analysis)
• In la ed in signi icance o low-exp ession genes
Addi ionally, he hallma k e ms we e no co ec ed o pa hway edundancy (e.g.,
o e lapping cell-cycle and p oli e a ion signa u es a i icially boos ing “ isk hallma ks”).
Ex ended Da a Figu e 7 – Mic oen i onmen quan i ica ion lacks
aw da a
The ib oblas ma ke quan i ica ions appea o use only ela i e in ensi ies, which can
be easily skewed by s aining ba ch di e ences. Raw mean pixel in ensi ies and he
no maliza ion me hod a e no p o ided.
Ex ended Da a Figu e 8 – Immune p o iling panel possibly o e -
in e p e ed
Flow cy ome y ga ing o immune sub ypes is unclea , and he ela ionship be ween
mamma y immune in il a ion and sys emic d ug exposu e is no es ed.
8
Ex ended Da a Figu e 9 – Missing nega i e con ols
The collagen emodeling expe imen s lack essen ial con ols such as:
• Placebo- ea ed mice
• Tamoxi en o a oma ase inhibi o a ms
• Vehicle-alone explan s
Wi hou hese, a ibu ing ECM changes o an i-p oges ins is specula i e.
Ex ended Da a Figu e 10 – Conclusions no suppo ed by
quan i ica ion
The summa y ca oon o e s a es he indings and claims “b oad supp ession o b eas
cance isk pa hways”, which is no igo ously suppo ed by he ea lie ex ended da a.
Ex ended Da a Figu e 11 – Lack o alida ion o p oges e one
ecep o downs eam a ge s
This igu e a emp s o alida e changes in canonical p oges e one ecep o (PR) a ge
genes ollowing an i-p oges in ea men . Howe e :
• The qPCR da a lack p ime e iciency cu es, mel ing cu es, o e i ica ion o
single-amplicon p oduc s.
• Se e al key PR- esponsi e genes (e.g., RANKL, WNT4) show la ge dono - o-dono
a iabili y ha unde mines s a is ical con idence.
• The au ho s ely on ela i e exp ession (ΔΔC ) wi hou showing aw C alues,
making i unclea whe he obse ed di e ences e lec ue biological changes o
noise nea he de ec ion limi .
Mo eo e , he ange o genes es ed seems selec i e, ocusing only on hose mo e likely
o change, which may in oduce con i ma ion bias.
Ex ended Da a Figu e 12 – Po en ial o e i ing o machine-
lea ning classi ie s
The s udy in oduces a classi ie o dis inguish ea ed s. un ea ed explan s based on
ansc ip ional signa u es. Howe e :
• The classi ie appea s o be ained and e alua ed on he same da ase , aising
conce ns abou o e i ing.
• The numbe o samples (n ≈ 20–30) is oo small o obus machine-lea ning
modeling.
• No c oss- alida ion, independen es se , o ex e nal da ase is used o e alua e
gene alizabili y.
9
Gi en hese me hodological weaknesses, he epo ed accu acy me ics likely o e s a e
he model’s p edic i e powe .
Ex ended Da a Figu e 13 – Incomple e s omal cell- ype
cha ac e iza ion
This igu e a emp s o cha ac e ize ib oblas and myoepi helial subse s using bulk o
single-cell RNA-seq. Howe e :
• The clus e ing esolu ion appea s oo high, gene a ing many small clus e s
lacking biological jus i ica ion.
• Ma ke gene exp ession is inconsis en ac oss eplica es, hin ing a ba ch e ec s
o sample deg ada ion.
• The au ho s do no benchma k hei s omal clus e s agains published
mamma y s omal a lases, which is essen ial o alida ion.
In addi ion, he absence o independen alida ion using immunos aining (IHC o IF)
weakens con idence in hese ansc ip omic-de ined s omal s a es.
Ex ended Da a Figu e 14 – Mo phological quan i ica ion is
unde powe ed
The igu e p esen s quan i ica ion o epi helial hickness, duc al lumen size, and side-
b anch densi y ac oss ea men g oups. Ye :
• Sample sizes a e unclea .
• Va iance is high and dis ibu ion non-no mal, ye au ho s apply pa ame ic
s a is ics wi hou jus i ica ion.
• No co ec ion o mul iple measu emen s wi hin he same animal is pe o med.
The s a is ical agili y indica es ha mo phological di e ences may no be as obus as
claimed.
Ex ended Da a Figu e 15 – Apop osis assays lack con ols and
aw da a
TUNEL assays a e sensi i e o:
• Tissue ixa ion
• Pe meabiliza ion
• Enzyma ic eac ion ime
• Au o luo escence
Ye he au ho s do no show:
16
Un il such imp o emen s a e made, he s udy’s cen al claims should be in e p e ed
wi h cau ion.
Re e ence
1 Simoes, B. M. e al. An i-p oges in he apy a ge s hallma ks o b eas cance isk.
Na u e (2025). h ps://doi.o g/10.1038/s41586-025-09684-7
