Spatiotemporal Mapping of Tumor Microenvironment Remodeling During Fiber-Optic Photothermal Therapy: A Multiparametric MRI Study in 4T1 Breast Cancer Xenografts.

1
Introduction
Breast cancer (BC) persists
as a leading global health burden,
representing both the most prevalent malignancy and a principal cause
of cancer-related mortality among women worldwide. In China, there were approximately 357,200 new female BC
cases and 75,000 deaths in 2022, constituting 15.59% and 7.94% of
total new cancer cases and deaths. And
the incidence rates among the world population demonstrates the sustained
epidemiological escalation. The diagnostic
challenge is compounded by asymptomatic early stage progression and
insufficient screening modalities, frequently resulting in advanced-stage
detection that correlates with therapeutic resistance. Contemporary
clinical management relies on molecular subtyping
−

to guide multimodal
interventions including radical surgery, ionizing radiotherapy, cytotoxic
chemotherapy, and endocrine modulation. Nevertheless, these conventional
approaches face fundamental limitations: tumor heterogeneity and occult
micrometastases, which impede complete surgical resection. Also, cancer
stem cell populations mediated radioresistance and chemotherapeutic
evasion.
,
Furthermore, hormonal interventions remain
restricted to receptor-positive subtypes.
Emerging thermal vulnerability
paradigms suggest malignant cells
exhibit differential sensitivity to hyperthermic stress. Localized thermal ablation induces programmed
cell death through mitochondrial apoptosis pathways and direct protein
denaturation. Capitalizing on this vulnerability, photothermal therapy
(PTT) has emerged as a spatially controlled intervention employing
light-absorbing nanoparticles to transduce near-infrared radiation
(NIR) into cytotoxic hyperthermia (42–48 °C).
,
This modality enables precision tumor eradication while preserving
peri-tumoral architecture, particularly advantageous for superficial
breast malignancies. However, transcutaneous energy delivery suffers
from significant photon scattering and nonspecific tissue heating.
To circumvent these limitations, optical fibers--originally developed
for telecommunication--have been explored for their therapeutic potential.
Owing to their flexibility, biocompatibility, and multifunctionality,
they have become increasingly favored in minimally invasive surgical
procedures. Advances in fiber-optic-based
sensing and treatment have led to significant breakthroughs, including
in situ therapy and light-controlled drug delivery for tumors such
as glioma and colorectal carcinoma.
−

Further innovations
have been focus on target-driven-structured nanoparticles, Zhang et
al.
,
enhanced the photodynamic effect through
nanoparticle surface modifications while ensuring stability via a
novel phenylamine-based topological design with an optimized donor–acceptor
(D-A) structure. Further functionalization with Angiopep-2, an oligopeptide
ligand, enabled blood-brain barrier penetration and glioblastoma-targeted
therapy. These innovative approaches hold considerable promise for
future therapeutic developments.
−

The tumor microenvironment
(TME) constitutes a complex ecosystem
that sustains and promotes malignant cell growth. This dynamic bidirectional
interplay between the TME and tumor cells drives aggressive phenotypic
behaviors that facilitate cancer progression. Notably, the vascular
architecture and oxygenation dynamics within the TME play a pivotal
role in modulating therapeutic resistance mechanisms, while simultaneously
influencing multiple physiological processes for tumor advancement, fostering angiogenic dysregulation and immunosuppressive
niche formation that undermine conventional therapies. Noninvasive
monitoring of these parameters through advanced imaging biomarkers
has therefore become imperative. Intravoxel incoherent motion (IVIM)
MRI decouples Brownian diffusion (D) from microcirculatory
perfusion (f, D*) through biexponential
signal modeling, providing quantitative
maps of cellular density and neovascularization.
,
Concurrently, blood oxygen level-dependent (BOLD) MRI detects paramagnetic
susceptibility shifts induced by deoxyhemoglobin accumulation, serving as a surrogate for tissue hypoxia. Clinical
validations demonstrate strong correlation between IVIM parameters
(f: r = 0.82; D*: r = 0.79) and histopathological microvessel density
in BC, while BOLD signal intensity inversely
correlates (r = −0.85) with [18F]­FMISO PET
hypoxic volumes in prostate cancer models. Emerging evidence further suggests these modalities may predict
radiation response as a potential predictive imaging biomarker.
,

Despite the promising preclinical results from fiber-optic
PTT, critical knowledge gaps persist
regarding its
acute bioeffects and TME modulation. This investigation establishes
an orthotopic BC animal model to characterize the 24-h pathophysiological
cascade following intervention. Our multimodal approach integrates:
(1) IVIM and BOLD MRI for longitudinal perfusion and hypoxia mapping;
(2) The Fiber-optic pH telemetry for real-time acidosis monitoring;
(3) Histopathological correlation through K
i-67 (proliferation), HIF-1α (hypoxia), and TUNEL (apoptosis)
biomarkers. By synergizing these techniques, we aim to achieve four-dimensional
characterization (3D spatial + temporal) of TME remodeling, enabling
comprehensive evaluation of therapeutic efficacy beyond conventional
volumetric assessments.

2
Materials and Methods
All animal experiments
conducted in this study were approved by
the Institutional Animal Ethics Committee of Jinan University, adhering
to the protocols outline in the Institutional Laboratory Animal Care
and Use Manual (Approval No. 20220923–01).
2.1
Cell Culture and Animal Model
4T1-breast
cancer cells were procured from the College of Life Science and Technology
at Jinan University. These cells were cultured in Dulbecco’s
Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum
(FBS) and maintained at 37 °C in a 5%CO2 atmosphere.
Seventy-five female Balb/c mice, aged 6–8 weeks and weighing
20–25 g, were purchased from Beijing HFK bioscience Co., Ltd.
The mice were housed in a specific pathogen free (SPF) environment
at the Laboratory Animal Center of Jinan University. Breast cancer
xenografts were established by subcutaneously injecting 0.1 mL of
4T1 breast cancer cell suspension (1 × 10∧6 cells/ml)
into the left fat pad of each mouse. Tumor growth was monitored using
vernier calipers, and the experiment commenced once the tumor volume
reached 100–200 mm. We initiated the experiment after the tumor
reached about 100–200 mm3, during the exponential
growth phase (Figure
).

2.2
Experimental Design
Thirty mice with
established breast cancer xenografts were randomly selected and divided
into two cohorts of 15 mice each. Anesthesia was induced via intraperitoneal
administration of pentobarbital (50–60 mL/kg). Optical fibers
coated with graphene were inserted into the tumor using a guiding
needle in both cohorts. The experimental group underwent PTT assisted
by optical fibers, followed by an initial MRI scan, with continuous
monitoring using infrared detectors. Tumors were treated using an
infrared generator operating at 0.5 W and 1.3 mA. The control group
underwent the standard procedure without PTT. All mice underwent T1-weighted
imaging (T1WI), T2-weighted imaging (T2WI), intravoxel incoherent
motion MRI (IVIM-MRI), and blood oxygen level-dependent MRI (BOLD-MRI)
scans before and at 0.5, 1.0, 2.0, 4.0, and 24 h post-treatment.

2.3
Fiber Optic Fluorescence pH Sensor Preparation
2.3.1
pH Sensing Sol–Gel Preparation
The precursor sol of ethyl triethoxysilane (ETEOS) was prepared
by mixing ETEOS, ethanol, and 0.1 mol/L HCl in a molar ratio of 1:6.25:0.007
and aged for 7 days. The GPTMS precursor sol was prepared by mixing
GPTMS, 1-methylimidazole, deionized water, and ethanol in a molar
ratio of 1:0.69:4:6.25 and aged for 4 h. The pH sensing sol–gel
was prepared by mixing the ETEOS precursor sol, GPTMS precursor sol,
HPTS-IP ethanol solution (10 mmol/L), and DABCO ethanol solution (10
mmol/L) in a molar ratio of 1:1:0.001:0.001. The final sol–gel
mixture was left at room temperature for 3 days before being coated
onto the fiber tip.

2.4
Fabrication of Fiber-Optic Fluorescent pH
Sensor
The tapered fiber was immersed in the sol–gel
mixture for 30 min to deposit a pH sensing sol gel film on its surface.
The fiber was then cured in a nitrogen-circulating oven at 140 °C
for 4 h, with a heating rate of 2 °C/min. The fabricated fiber-optic
pH sensor was stored in a dark environment at 4 °C until use.

2.5
Instrumentation and Experimental Setup
The fiber-optic pH sensor was connected to an excitation light source
via a Y-shaped fiber-optic cable, utilizing 450 and 375 nm fiber coupled
lasers (customized by Innova Optoelectronics Technology Co., Ltd.
in Shenzhen). The sensor employed dual-excitation ratio fluorescence
detection by alternately exciting the indicator dye on the fiber surface
with 450 and 375 nm light. (Figure
C)

2.6
MRI Examinations
All MRI scans were
performed using a GE Discovery MR750 3.0 T scanner equipped with a
2-turn 12 mm coil for mice. T1-weighted images (T1WI) were acquired
using fast spin–echo (FSE) sequences with the following parameters:
repetition time (TR) = 625.0 ms, slice thickness = 2.9 mm, field of
view (FOV) = 30 × 30 mm2, matrix size = 288 ×
288, and number of excitations (NEX) = 2.0. T2-weighted images (T2WI)
were acquired using fast recovery fast spin–echo sequences
with TR = 3000.0 ms, TE = 68.0 ms, slice thickness = 15.63 mm, FOV
= 288 × 288, matrix size = 288 × 288, NEX = 4. IVIM-DWI
was acquired using an echo-planar imaging pulse sequence with TR =
1200.0 ms, TE = 102.4 ms, slice thickness = 2.9 mm, matrix size =
128 × 64, and FOV = 100 × 100 mm2. Twelve b-values were adopted: 102, 302, 502, 1002, 1504, 2004, 3004, 4004, 5004, 8004, 10006 s/mm
2
, with NEX = 4 for each b-value. BOLD-MRI images were acquried using a multiecho
FSPGR sequence with TR = 500 ms, TE = 38.5 ms, slice thickness= 2.2
mm, FOV = 8 × 8 cm, matrix size = 160 × 160, NEX = 2.

2.7
Image Analysis
MRI data was processed
using the Advantage Workstation Version 4.5 (ADW4.5, GE Healthcare)
postprocessing workstation. Two senior radiologists with 5 and 15
years of experience analyzed the data, with the latter repeating the
analysis one month later. Each tumor was divided into three regions
based on the distance from the center of the PTT area: the center
area, peripheral area, and margin area. The biexponantial model of
IVIM-DWI was expressed aswhere S
0 is the
mean signal intansity, Sb
is signal intansity
at high b-values. D is the true
diffusion coefficient, D* is the pseudodiffusion
coefficient, and f is the perfusion fraction.
The BOLD-MRI model under a gradient echo magnetic field
expressed aswhere ΔB
inhom represents the heterogeneity within the matrix, and γ is the
gyromagnetic ratio of proton. The transverse relaxation rate was used
to calculate the signal attenuation ratio
Where S
0 is the
initial signal intensity, and TE is the echo time.

2.8
Histological and Immunohistochemical Analysis
Thirty-six of 75 tumor-bearing mice were chosen for pathological
analysis. Three mice from each group (experimental and control) were
randomly assigned to and each time point. The mice were euthanized
via pentobarbital overdose, and the tumors were excised, fixed in
10% paraformaldehyde, embedded in paraffin, and sectioned at 3 μm
thickness. Sections were stained with hematoxylin and eosin (H&E)
following standard protocols. K-67 staining was performed
using an anti-K
i-67 antibody (1:1000;
Abcam). Hypoxia-inducible factor-1alpha (HIF-1α) staining was
performed using a monoclonal anti HIF-1α antibody (1:100; Affinity,
#AF1009). Termial deoxynucleoitidyl transferase mediated dUTP nick
end labeling (TUNEL) immunoflurescent staining was performed using
a TUNEL kit (Novelbio, China) accroding to the manufacturer’s
instructions.
Three random fields from each section were captured
using an optical microscope, and the average value of these fields
was considered the final value of that section. All sections were
analyzed using Aipathwell Software. For K
i-67 and HIF-1α stain, the average integrated optical density
(IOD) was calculated for each high-magnification (×200) field
of view using the formula

For TUNEL-stained sections, the percentage
of positively stained
cells was calculated as

2.9
Statistical Analysis
Statistical
analysis was conducted using SPSS 27.0 software, and graphs were gnerated
using using Prsim 10.0 software (GraphPAd Software Inc., San Diego,
CA). Quantitative MRI and pH data, as well as semiquantitative pathological
data, were expressed as mean ± standard deviation (X ± S). The homogeneity of quantitative data
was assessed using the Kolmogorov–Smirnov test. Repeated measures
analysis of variance (ANOVA with LSD post hoc) was used to compare
imaging parameters and pathological indicators between experimental
and control groups at different time points. Pearson correlation was
used to assess the relationship between MRI parameters and pathological
indicators. A p value <0.05 was considered statistically
significant. In corrrlation analysis, an r value
≥ 0.75 indicated a very strong correlation, 0.5 ≤ r < 0.75 indicated a moderate to good correlation. 0.25
≤ r < 0.5 idicated an average correlation.
and r < 0.25 indicated a very weak correlation.

3
Result
3.1
Intra- and Interobserver Agreement on IVIM
and BOLD-fMRI Parameters
The intra- and interobserver reliability
for IVIM and BOLD parameter measurements demonstrated good-to-excellent
agreement in both experimental and control groups, as quantified by
intraclass correlation coefficients (ICCs). Complete ICC values (ranging
from 0.77 to 0.93) are detailed in Supplementary Fold.

3.2
Fiber Optic-Assisted Photothermal Therapy
for Tumor Treatment
Infrared imaging was conducted during
the treatment process. At the initiation of the therapy, the tumor
area temperature was slightly below body temperature. Following the
commencement of photothermal therapy (PTT), the local tumor temperature
increased, forming a distinct bright spot on the infrared imaging,
with the central temperature reaching 50–55 °C (Figure
A). Shown in Figure
B, the tumor region
is divided into three areas by two equidistant points marked along
a straight line extending from the treatment location to the tumor
edge.

3.3
Comparison of IVIM and BOLD Parameters between
the Control and Experiment Groups
The mean values of IVIM-DWI
and BOLD parameters for both experimental and control groups at each
time point are summarized in Figure
and Tables
, and and
visualized in Figure
. Quantitative MRI parameters, including D, D*, f, and R2* values,
exhibited distinct trends across different tumor regions in both groups.
3.3.1
D Values
The intergroup
comparison of D values at each scanning time point
is illustrated in Figure
(A–C). At 0.5, 1, 2, and 24 h post-treatment, the D values in the central and peripheral tumor regions of
the experimental group were significantly higher than those in the
control group (p < 0.05). At 4.0 h, the D values in the experimental group decreased to levels comparable
to the control group, with no significant difference (p > 0.05), but subsequently increased again, reaching higher levels
by the observation end point (24 h). No significant differences in D values were observed in the marginal zone between the
experimental and control groups at any time point (p > 0.05).1.

Dand

f

values*: Both D* and f values exhibited an initial increase post-treatment,
significantly higher than those in the control group (p < 0.05), followed by a steady decline, reaching their lowest
levels at the 24-h time point (p < 0.05). The
decline in f values was more pronounced (p < 0.001) (Figure
(D–I)).

3.3.2
R2* Values
The R2* values in the central tumor region initially decreased,
followed by fluctuations, and reached their lowest point at 24 h.
In the peripheral region, the R2* values remained
relatively stable during the 0–4 h time points but also declined
to their lowest point by 24 h. No significant differences in R2* values were observed in the marginal regions at any
time point (Figure
(J–L)).

3.4
Fiber Optic Fluorescence pH Sensor Based On
Hydrogel For Tumor pH Detection
As shown in Figure
, the pH values of tumors in
various regions of the experimental group significantly increased
after PTT compared to the control group (p < 0.05).
In contrast, the control group exhibited no significant changes in
pH values following treatment.

3.5
Pathological Analysis
Histological
examinations using H&E staining demonstrate densely packed and
disodered tumor cells with varying nuclear morphology at baseline
of both groups (Figure
). Following fiber optic-assisted PTT in the experimental group,
large-scale cell necrosis was observed, characterized by shrunken,
fragmented nuclei scattered within the cytoplasm. TUNEL staining positivity
in the center area (Tables
– and Figure
­(G–I)) in the
experimental group increased dramatically, particularly at 24-h time
point (P < 0.001). K
i-67 expressions in center and peripheral area (Tables
– and Figure
(A–C)) showed
a noticeable decline during the first 4 h, followed by a slight increase
at 24 h. No significant differences in K
i-67 expression were observed in the marginal region between experimental
group and control group. These findings suggest that the PTT exerts
an antitumor effect. HIF-1α expression followed a similar trend
to K
i-67, exhibiting an initial decline
post-therapy, followed by a rebound, which was more pronounced in
the tumor margin (Tables
– and Figure
(D–F). Together, these findings and
the representative histopathological images (Figure
) collectively demonstrate the treatment-induced
pathological alterations.

3.6
Correlation of Histologic Features With IVIM
and BOLD Parameters
The results of the semiquantitative correlation
analysis between imaging parameters and pathological features are
shown in Figure
.
3.6.1
Apoptosis and Proliferation Activity
The f value showed a very strong correlation with K
i-67 (r = 0.7333, P =
0.0019). Moderate correlations were observed between D* and K
i-67 (r = 0.594, P = 0.012) and between R2* and K
i-67 (r = 0.543, p = 0.036). The D value exhibited a strong negative
correlation with K
i-67 when considered
only baseline and 24-h time points, however, no significant correlation
was observed between D and K
i-67 over
the entire observation period (Figure
).

3.6.2
TUNEL Correlation

R2* was negatively correlated with TUNEL (r = −0.748, p < 0.001), while f and D* values of
IVIM correlated with TUNEL (f-IVIM: r = 0.595, p =
0.003; D*-IVIM: r = −0.560, p = 0.013).

3.6.3
Tumor Hypoxia

R2* showed a significant correlation with HIF-1α (r = 0.7769, p = 0.001), and the f value of IVIM also correlated with HIF-1α (r = 0.3879, p = 0.0173).

4
Discussion
Malignant tumors represent
a global health crisis, demanding innovative
therapeutic strategies. Fiber-optic PTT has emerged as a promising
modality due to its precision, real-time monitoring capacity, and
minimally invasive nature. While existing studies primarily focus
on macroscopic therapeutic outcomes, this
investigation bridges a critical knowledge gap by systematically characterizing
early microenvironmental alterations using multiparametric MRI, fiber-optic
pH sensing, and histopathological analyses in a murine 4T1 breast
cancer model.
4.1
Therapeutic Efficacy and Immediate Effects
Our thermographic data demonstrate controlled temperature elevation
from 30 to 55 °C within the subcutaneous tumors during PTT, achieving
the optimal range for apoptotic induction while minimizing metastasis
risks associated with hyperthermia,. The
immediate tissue carbonization observed at treatment sites underscores
the need for refined thermal distribution control in future device
iterations. Histopathological validation revealed dual mechanisms
of action: (1) direct thermal ablation evidenced by H&E-stained
nuclear fragmentation and (2) programmed cell death confirmed through
TUNEL assay, with apoptosis peaking at 24 h post-treatment. Concordantly, K
i-67 suppression indicated sustained antiproliferative
effects, aligning with previous reports on PTT-mediated cell cycle
arrest.
,

4.2
Tumor Microenvironment Remodeling
IVIM-DWI revealed dynamic spatiotemporal changes in tumor biophysics.
The biphasic D-value trajectoryinitial surge
(cellular membrane disruption), subsequent decline (tissue condensation)
and late-phase recovery (cellular disintegration)mirrors the
temporal evolution of the PTT-induced microstructural modifications.
Our result is in agreement with previous thermal ablation study which
suggest that the initial increase in D value due to a temporal raising
temperature and the following decrease occurred during the heat dissipated,
necrosis factor overrode the temperature-based increases.
,
However, it is not clear how this information could be separated
from the temperature-based increases in ADC while the tissue is heated.
While IVIM-DWI detects microstructural changes, its clinical translation
is challenged by parameter nonspecificity. The observed D-value fluctuations (initial rise: +26.7% at 1 h) likely reflect
competing effects of temperature-enhanced water mobility and cellular disruption, necessitating future
histology-matched voxel analysis. The heating situation weakens the
correlation between D value from IVIM and immunohistochemical
assays, especially K
i-67, which found
correlated in other studies.
,
By the 24-h time point,
the D value increased again, consistent with findings
from recent thermal ablation and PTT studies,
,,−

as disrupted cell membranes
facilitated enhanced water diffusion, ADC values increased from below
normal to above normal. Notably, while D-values showed
limited correlation with proliferation/apoptosis markers (K
i-67: r = −0.3893, p = 0.335; TUNEL: r = 0.086, p = 0.691), the strong association between D*/f parameters and vascular markers establishes these IVIM indices as robust early predictors of microcirculatory
changes. The transient D*/f elevation
(inflammatory vasodilation) followed by progressive decline (vascular
necrosis) suggests PTT initiates vascular normalization preceding
structural collapse. According to previous studies, this decrease
is associated with tissue necrosis during treatment, which disrupts
microvascular components in the affected regions. The destruction
of the capillary network in the treatment area resulted in reduced
perfusion and lower f values.
−

4.2.1
Hypoxia Modulation and Metabolic Shifts
BOLD-MRI detected progressive R2* reduction, particularly
in central tumor regions, paralleling fiber-optic pH sensor measurements
of alkalization (Figure
). This triphasic hypoxia alleviation(1) reduced oxygen consumption
via cell death, (2) improved perfusion through vascular leakage reduction,
and (3) Warburg effect attenuation was further corroborated by HIF-1α
downregulation (Figure
). The spatial-temporal discordance in R2* responses
(central: immediate decrease vs peripheral: 24 h delay) likely reflects
inherent oxygen gradients in 4T1 breast cancer model. And oxygenated
peripheries exhibited delayed therapeutic susceptibility. Previous
research has shown that the decline in R2* values
is associated with tumor regeneration and increased blood perfusion,
leading to improved oxygenation. Cao
et al. used BOLD to monitor tumor changes
following PTT and photodynamic therapy, observing a continuous decline
in R2* values over 10 days on using photothermal
therapy alone. However, the observed R2* reduction
post-PTT could represent either therapeutic oxygenation improvement
or vascular collapse reducing deoxyhemoglobin delivery. This ambiguity highlights the need for complementary
techniques.

4.3
Regional Heterogeneity and Clinical Implications
Spatial analysis uncovered distinct treatment responses across
tumor subregions. While central and peripheral zones showed synchronized
IVIM parameter trajectories, margin areas demonstrated delayed apoptosis
(24 h TUNEL surge) and transient HIF-1α suppression followed
by rebound. This spatial hierarchy suggests: (1) Central-peripheral
regions: Primary ablation zones with immediate vascular/metabolic
effects. (2) Margin regions: Secondary therapeutic targets via heat
diffusion and delayed apoptotic cascades. These findings emphasize
the necessity for 3D treatment planning to ensure complete coverage
of tumor extensions.

4.4
Technical Advantages and Limitations
Our integrated approach synergizes fiber-optic thermometry with multiparametric
MRI, enabling real-time therapy guidance and quantitative outcome
assessment. The differential utility of MRI biomarkersD*/f for early response, D for long-term
monitoring, R2* for hypoxia evaluationprovides
a framework for personalized treatment optimization. However, limitations
include: (1) Sample size constraints reducing statistical power; (2)
Subcutaneous model limitations in replicating human organ microenvironments;
(3) 24 h observation window potentially missing late-phase changes;
(4) Fourth, this study acknowledges inherent technical limitations
of the imaging modalities employed. The IVIM model, while valuable,
is sensitive to confounding factors. The derived parameters, particularly
the perfusion fraction (f) and pseudodiffusion coefficient
(
D
*), can be influenced by restricted
diffusion in dense tissue and are inherently sensitive to temperature
changes, such as those induced by PTT itself, which may complicate
a pure interpretation of microcirculatory changes.
,
Similarly, BOLD contrast is a nonspecific measure of deoxyhemoglobin
concentration. The observed reduction in R2*, while
suggestive of improved oxygenation, could also theoretically result
from a collapse of the vascular network and a consequent reduction
in deoxyhemoglobin delivery, highlighting the need for complementary
techniques to confirm hemodynamic improvements.

5
Conclusion
In conclusion, fiber-optic
PTT exerts multidimensional effects
via coordinated mechanisms: direct cytotoxicity (apoptosis/necrosis),
microvascular modulation (D*/f changes),
and hypoxia alleviation (R2* reduction). IVIM and
BOLD parameters not only serve as early biomarkers but also reflect
the temporal sequence of therapeutic effects, advancing personalized
BC therapy optimization.

Supplementary Material