Weijian Wang1, Yimeng Kang1, Wenjing Li1, Shujian Li1, Wenhua Zhang1, Kun Zhang1, Liangjie Lin2, Zhigang Wu2, Peng Sun2, Yong Zhang3, and Jingliang Cheng1
1MRI, The First Affiliated Hospital of Zhengzhou University, Zhengzhou, Zhengzhou, China, 2Advanced Technical Support, Philips Healthcare, Beijing, China, 3the First Affiliated Hospital of Zhengzhou University, Zhengzhou, China
Synopsis
Keywords: Microstructure, Diffusion/other diffusion imaging techniques, cervical cancer, immunohistochemistry, time-dependent diffusion MRI, microstructural mapping, oscillating gradient spin-echo
Motivation: Some immunohistochemical (IHC) markers are gradually accepted as vital prognostic factors guiding therapy in cervical cancer (CC).
Goal(s): This study aimed to evaluate the relationship of time-dependent diffusion MRI (td-dMRI)-based microstructural mapping and IHC status of CC.
Approach: Quantitative information on cell microstructure are obtained by fitting the IMPULSED model to the td-dMRI data.
Results: Intracellular fraction (fin) and cellularity were significantly correlated with the PD-L1 expression, respectively. In addition, fin and cellularity were significantly correlated with Ki-67. The results suggested that microstructural mapping might facilitate the establishment of novel diffusion-derived biomarkers to estimate IHC status and orient treatment of CC.
Impact: The td-dMRI-based microstructural mapping showed the
potential to estimate the IHC status of cervical cancer for the first time.
Introduction
Recently, numerous immunohistochemical (IHC)
biomarkers associated with hallmarks of cervical cancer and prognosis have been
identified. The Ki-67 is biomarker associated with proliferation, the human
epidermal growth factor receptor 2 (HER2) is associated with tumourigenesis and
apoptosis, and the biomarker p53 is associated with apoptosis. Evasion of
immune destructing has been recognised as a new hallmark of cancer, and the IHC
biomarker programmed death ligand 1 (PD-L1) is involved in immune escape
pathways. Pembrolizumab, a programmed death-1 receptor inhibitor, is widely
investigated and has received US Food and Drug Administration approval for
patients with persistent, recurrent, or metastatic cervical cancer with a PD-L1
combined positive score (CPS) of ≥1 based on Keynote-158 [1]. This study aimed to
investigate whether the tumor microstructural properties obtained from
diffusion time-dependent diffusion MRI (td-dMRI)
could be used to reflect the expression of prognostic IHC biomarkers in
cervical cancer.Methods
Seventy-one
patients with pathologically confirmed cervical cancer underwent MRI on a 3T
system (Ingenia Elition, Philips Healthcare, Best, the Netherlands). A
house-made oscillating gradient spin-echo (OGSE) dMRI sequence with
trapezoid-cosine gradients and pulse gradient spin-echo (PGSE) dMRI sequence
were implemented with 2D echo-planar imaging acquisition. OGSE data were
acquired at 17 Hz (effective td = 15 ms, 1 cycle, b =
0/250/500/750/1000 s/mm2), and 33 Hz (effective td =
7.5 ms, 2 cycles, b = 0/100/200/300 s/mm2). PGSE was acquired
with b-value of 0/250/500/750/1000/1400/1800 s/mm2. Quantitative
information on cell microstructure are obtained by fitting the IMPULSED
(Imaging Microstructural Parameters Using Limited Spectrally Edited Diffusion)
model to the td-dMRI data [2]. The microstructural parameters
included mean cell diameter (d), intracellular fraction (fin), and
extracellular diffusivity (Dex), while intracellular diffusivity (Din)
was fixed at 1.58 μm2/ms to ensure fitting stability. Cellularity
was represented as fin/d for simplicity. In addition, apparent
diffusion coefficient (ADC) was calculated with all b values at each td.
The regions-of-interest (ROIs) of tumor
tissue were manually delineated on a representative single slice of DW images
showing the maximum tumor area. The necrotic area or surrounding tissue was
carefully excluded from the segmentation. The averaged values of microstructural
parameters within the tumor ROIs (d, fin, Dex, and ADC)
were recorded. Spearman
correlation was applied to evaluate the association between MR parameters and PD-L1
expression, parameters
and Ki-67 LI. Parameter differences were assessed between different HER2 and
P53 status.Results
The Spearman correlation analysis showed that fin and cellularity
were significantly correlated with the PD-L1 expression (r = 0.332 and 0.358, respectively, both p < 0.05). ADC values for OGSE/17Hz
(ADC17Hz), and PGSE (ADCPGSE) were negatively correlated
with the PD-L1 expression (r = -0.361 and 0.-369, respectively, both p < 0.05). No statistically significant associations
were found between other parameters (d, Dex, ADC33Hz)
and PD-L1 expression. In addition, fin and cellularity were significantly
correlated with Ki-67 LI (r = 0.566 and 0.442, respectively, both p < 0.05). The d, Dex, ADC17Hz, ADC33Hz
and ADCPGSE had no significant correlation with Ki-67 LI. None
of the parameters differ significantly between HER2-positive and HER2-negaitive
tumor, between P53-positive and P53-negaitive tumor. Representative images from
conventional T2W, ADC, and td-dMRI-based
microstructural mapping are shown in Figure 1.Discussion
In this study, our results demonstrated that both td-dMRI-based microstructural parameters and conventional ADCPGSE
can be used to reflect the expression of PD-L1 in cervical cancer. It is
reported that poorly differentiated cervical cancer had higher PD-L1 expression
compared with grade 2 tumors [3]. The higher cell density and more significant
cellular atypia in poorly differentiated tumor may result in the increased cellularity,
intracellular fraction and decreased ADC. In addition, high PD-L1 levels was
associated with high CD8 positive lymphocytic infiltrates in cervical cancer
[3], which may also increased the cellularity in tumor tissue. Our study found that
the microstructural parameters had advantages over the ADCPGSE
in the prediction of Ki-67 LI. This result may indicate that td-dMRI-based microstructural mapping may be better than conventional PGSE dMRI in
reflecting cell proliferation level in cervical cancer. OGSE dMRI can shorten
the diffusion times by using rapidly oscillating gradients and is sensitive to
smaller spatial scales in the tissues compared with conventional PGSE dMRI [4].
By fitting a combination of OGSE and PGSE signals to specific biophysical
models, we can estimate important microstructural properties such as cell size,
cell volume fraction, and cellularity, which are closely related to the
pathological changes of tumor.Conclusion
Our study demonstrated the feasibility of
reflecting prognostic IHC biomarkers status in cervical cancer using the
td-dMRI-based microstructural parameters for the first time.Acknowledgements
NoReferences
- Chung HC, Ros W, Delord JP, et
al. Efficacy and Safety
of Pembrolizumab in Previously Treated Advanced Cervical Cancer: Results From
the Phase II KEYNOTE-158 Study. J Clin Oncol 2019;37:1470-1478.
-
Xu
J, Jiang X, Li H, et al. Magnetic resonance imaging of mean cell size in human
breast tumors. Magn Reson Med 2020;83:2002-2014.
- Saglam
O, Zhou J, Wang X, et al. PD-L1 Expression Correlates With Young Age and CD8+
TIL Density in Poorly Differentiated Cervical Squamous Cell Carcinoma. Int J
Gynecol Pathol 2020;39:428-435.
- Ejima F, Fukukura Y, Kamimura K, et
al. Oscillating Gradient
Diffusion-Weighted MRI for Risk Stratification of Uterine Endometrial Cancer. J
Magn Reson Imaging 2023; Epub ahead of print. PMID: 37886909.