Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
Neoplasia 44 (2023) 100934
1476-5586/© 2023 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
Original Research
Increased expression of IDO1 is associated with improved survival and
increased number of TILs in patients with high-grade serous ovarian cancer
Inga Hoffmann
a
, Mihnea P. Dragomir
a
,
b
,
c
, Nanna Monj´
e
a
, Carlotta Keunecke
d
,
Catarina Alisa Kunze
a
, Simon Schallenberg
a
, Sofya Marchenko
a
, Wolfgang D. Schmitt
a
,
Hagen Kulbe
d
,
e
, Jalid Sehouli
d
,
e
, Ioana Elena Braicu
d
,
e
, Paul Jank
f
, Carsten Denkert
f
,
Silvia Darb-Esfahani
g
, David Horst
a
, Bruno V. Sinn
a
, Christine Sers
a
, Philip Bischoff
a
,
b
,
c
,
1
,
Eliane T. Taube
a
,
1
,
*
a
Institute of Pathology, Charit´
e – Universit¨
atsmedizin Berlin, corporate member of Freie Universit¨
at Berlin and Humboldt Universit¨
at zu Berlin, Charit´
eplatz 1, 10117
Berlin, Germany
b
Berlin Institute of Health at Charit´
e - Universit¨
atsmedizin Berlin, Charit´
eplatz 1, 10117 Berlin, Germany
c
German Cancer Consortium (DKTK), Partner Site Berlin, and German Cancer Research Center (DKFZ), 69120 Heidelberg, Germany
d
Department of Gynecology, European Competence Center for Ovarian Cancer, Charit´
e - Universit¨
atsmedizin Berlin, corporate member of Freie Universit¨
at Berlin,
Humboldt-Universit¨
at zu Berlin, and Berlin Institute of Health, 10117 Berlin, Germany
e
Tumorbank Ovarian Cancer Network, Charit´
e - Universit¨
atsmedizin Berlin, corporate member of Freie Universit¨
at Berlin, Humboldt-Universit¨
at zu Berlin, and Berlin
Institute of Health, 10117 Berlin, Germany
f
Institute of Pathology, Philipps-University Marburg and University Hospital Marburg, Baldingerstraße, 35043 Marburg, Germany
g
MVZ Pathologie Spandau, Stadtrandstr. 555, 13589 Berlin Spandau; MVZ Pathologie Berlin-Buch, Lindenberger Weg 27, Haus 207, 13125 Berlin
ARTICLE INFO
Keywords:
HGSOC
TILs
IDO1
Ovarian cancer
Prognostic marker
ABSTRACT
Background: The enzyme indoleamine 2,3-dioxygenase 1 (IDO1) plays a crucial role in regulating the immune
system’s response to tumors, but its exact role in cancer, especially in high-grade serous ovarian cancer (HGSOC),
remains controversial. We aimed to investigate the prognostic impact of IDO1 expression and its correlation with
tumor-inltrating lymphocytes (TILs) in HGSOC.
Methods: Immunohistochemical (IHC) staining and bioimage analysis using the QuPath software were employed
to assess IDO1 protein expression in a well-characterized cohort of 507 patients with primary HGSOC. Statistical
evaluation was performed using SPSS, and in silico validation considering IDO1 mRNA expression in bulk and
single-cell gene expression datasets was conducted. Additionally, IDO1 expression in interferon-gamma (IFNG)
stimulated HGSOC cell lines was analyzed.
Results: Our ndings revealed that IDO1 protein and mRNA expression serve as positive prognostic markers for
overall survival (OS) and progression-free survival (PFS) in HGSOC. High IDO1 expression was associated with a
signicant improvement in OS by 21 months (p <0.001) and PFS by 6 months (p =0.016). Notably, elevated
IDO1 expression correlated with an increased number of CD3+(p <0.001), CD4+(p <0.001), and CD8+TILs
(p <0.001). Furthermore, high IDO1 mRNA expression and protein level were found to be associated with
enhanced responsiveness to pro-inammatory cytokines, particularly IFNG.
Conclusions: Our study provides evidence that IDO1 expression serves as a positive prognostic marker in HGSOC
and is associated with an increased number of CD3+, CD4+and CD8+TILs. Understanding the intricate rela-
tionship between IDO1, TILs, and the tumor microenvironment may hold the key to improving outcomes in
HGSOC.
* Corresponding author.
E-mail address: eliane.taube@charite.de (E.T. Taube).
1
Contributed equally to this work.
Contents lists available at ScienceDirect
Neoplasia
journal homepage: www.elsevier.com/locate/neo
https://doi.org/10.1016/j.neo.2023.100934
Received 22 June 2023; Accepted 31 August 2023
Neoplasia 44 (2023) 100934
2
Introduction
With 313,959 new cases and 207,252 new deaths in 2020, ovarian
cancer is the 8th most common cancer-related death in females world-
wide [1]. High-grade serous ovarian carcinoma (HGSOC) constitutes the
most frequent histological subtype and is often diagnosed in advanced
stage [2]. With a 5-year survival rate of 43%, the prognosis of this dis-
ease is very poor and novel therapeutic strategies are urgently required
[3].
Cancer immunotherapy has shown efcacy in various types of can-
cer, such as melanoma, lung cancer, and bladder cancer. This thera-
peutic approach mainly comprises immune checkpoint inhibition
targeting the programmed cell death protein 1 (PD-1)/programmed
death-ligand 1 (PD-L1) pathway. However, inhibition of PD-1/PD-L1
has shown low efcacy in HGSOC [4]. Besides PD-1 and PD-L1, other
immune checkpoints might have to be targeted to overcome intrinsic
immunotherapy resistance that characterizes HGSOC.
One of these immune checkpoints is indoleamine 2,3-dioxygenase 1
(IDO1), an intracellular enzyme that plays a key role in regulating the
immune system’s response to invading pathogens and tumors. IDO1
expression can be modulated by a variety of signals, including growth
factors and pro-inammatory cytokines, such as interferon-gamma
(IFNG) that is secreted in large quantities during anti-tumor response.
In fact, IDO1 was the rst gene described to be activated by IFNG [5].
IDO1 catalyzes the initial and rate-limiting step in tryptophan
degradation, thereby limiting the availability of this essential amino
acid to cells and thus modulating their function and survival. In healthy
conditions, IDO1 balances the immune system by suppressing the im-
mune response through inhibiting T cell activation and promoting reg-
ulatory T cell recruitment. In numerous types of cancer, IDO1 has been
shown to be overexpressed [6]. While some studies demonstrate
immunosuppressive effects of IDO1, indicated by reduced number of
tumor-inltrating lymphocytes (TILs) and worse survival [7,8], other
studies found IDO1 correlated with higher TILs and better survival
[9–11]. In HGSOC, existing studies report a negative correlation be-
tween IDO1 expression and survival, but these studies are limited by
small cohort sizes [12–14]. IDO1 is considered a potential new target for
cancer immunotherapy [15,16], but its role in HGSOC remains unclear.
Due to the controversial ndings on the role of IDO1 in predicting cancer
prognosis and its suitability for therapeutic approaches, we reassessed
its impact on HGSOC in a well characterized large cohort with well
documented follow-up. We immunohistochemically investigated IDO1
protein expression patterns in a cohort of 507 patients correlating the
results with survival data. Additionally, we validated our results using in
silico analyses of IDO1 mRNA expression levels from publicly available
data sets. Finally, we performed exploratory biological studies using
RNA single cell analysis and in vitro HGSOC cell data.
Materials and methods
Study population and histopathological examination
The study cohort included tumor tissue samples of 507 patients
diagnosed with high-grade serous ovarian carcinoma (HGSOC). All pa-
tients underwent cytoreductive surgery at the Department of Gynae-
cology, Charit´
e - Universit¨
atsmedizin Berlin, Germany between 2000
and 2019. The samples were examined by two board-approved gyne-
cological pathologists (E.T., S.D.-E.) at the Institute of Pathology at
Charit´
e - Universit¨
atsmedizin Berlin, Germany and histological subtypes
were re-evaluated in accordance with the current WHO criteria [17]. As
quality control, all samples were tested for mutational p53 expression by
IHC. For the entire cohort, data on overall survival (OS) were available,
while data on progression-free survival (PFS) were available for 333
(66.7%) of these patients. The Charit´
e Comprehensive Cancer Center
(https://cccc.charite.de) and the Tumor Bank Ovarian Cancer Network
(www.toc-network.de) were used to obtain clinical data of the included
patients. A detailed overview of the study cohort characteristics can be
found in Supplementary Table 1. This study was performed in accor-
dance with the Declaration of Helsinki and the guidelines of the local
ethics committee (EA1/110/22).
Immunohistochemical staining
For IHC, tissue microarrays (TMAs) were prepared from formalin-
xed and parafn-embedded (FFPE) tissues of the HGSOC patients.
Briey, two cores of 1.0 to 1.5mm diameter for each patient were
punched out and assembled on 20 TMA blocks. The staining process was
performed on the DISCOVERY ULTRA autostainer (Ventana Medical
System, Inc., Tucson, Arizona, USA) with previous establishment of the
IDO1 antibody staining on normal tissue based on manufacturer’s in-
structions. For immunohistochemical IDO1 staining, tissue sections
were incubated with a monoclonal rabbit anti-IDO1 antibody (Ventana,
#86630) at 1:100 dilution after a Heat Induced Epitope Retrieval (HIER)
in standard CC1 Tris-EDTA buffer. Tissue of palatine tonsils served as
positive controls, while liver tissue was used as a negative control.
Image analysis
To perform digital image analysis, the stained TMA slides were
digitized using the Pannoramic Slide Scanner (3D Histech, Budapest,
Hungary) and subsequently analyzed with the open-source software
platform QuPath (version 0.3.0, [18]). A detailed description of the
procedure can be found in the Supplementary Materials, detailed
QuPath analysis parameters are listed in the Supplementary Tables S2
and S3.
In silico validation of survival analysis
To explore the expression of IDO1 on mRNA level and its effects on
survival, we used the online Kaplan-Meier plotter that enables the
genome-wide validation of several biomarkers in different cancer en-
tities [19,20]. The Kaplan-Meier plotter calculates an auto-selected
cut-off based on the best possible threshold to divide the cohort into
two groups and generates survival curves for them. Since HGSOC could
not be selected as a specic entity, we examined all serous carcinomas of
grade 2 and 3 together for OS and PFS in all available data sets
(GSE14764, GSE15622, GSE18520, GSE19829, GSE23554, GSE26193,
GSE26712, GSE27651, GSE30161, GSE3194, GSE51373, GSE63885,
GSE65986, GSE9891, TCGA).
In silico bulk gene expression analysis
RSEM-normalized gene expression data of the HGSOC cohort [21,22]
of The Cancer Genome Atlas (TCGA) was downloaded from cBioPortal
(https://www.cbioportal.org/, accessed on Nov 5, 2022). Gene expres-
sion data was available for 300 cases. Spearman correlation coefcients
were calculated for all genes. Genes were ranked by correlation with
IDO1 expression. Based on the ranked gene list, gene set enrichment
analysis (GSEA) was performed using the “clusterProler” R package
[23]. The code used for TCGA bulk gene expression analysis is available
from: https://github.com/bischofp/HGSOC_IDO1.
In silico single-cell gene expression analysis
Single-cell gene expression data and metadata were downloaded
from two public datasets [24,25]. Analyses were performed using the
open-source software “R” (version 4.1.1) and package “Seurat” (version
4.1.0; [26]) and the gene signatures Hallmark [27], KEGG [28], as well
as two antigen processing and presentation signatures [29,30]. For
detailed description of all analysis parameters, see Supplementary
Materials.
I. Hoffmann et al.
Neoplasia 44 (2023) 100934
3
Cell culture and interferon-gamma stimulation
OAW-42, OVCAR-3, and SKOV-3 cells were cultured in DMEM
(Gibco #21885-025) and supplemented with 10% fetal bovine serum,
no added antibiotics, at 37◦C with 5% CO2 and 95% humidity. Prior to
the study, cytogenetic analysis and cell authentication was performed at
the DNA-Fingerprinting Facility at Charit´
e Berlin using short tandem
repeat DNA. All cell lines were tested for mycoplasma contamination
using PCR Mykoplasmen kit (Biontex #M030/050). Cells were plated in
10cm dishes and when a 70% conuence was reached, they were
stimulated with human IFNG (Sigma-Aldrich #SRP3058), 1000U/ml.
After 24 hours, the cells were harvested for subsequent experiments. As
controls, we used unstimulated cell lines undergoing the same plating
and incubation protocol. Western blotting was performed subsequently
according to established standards at the Institute of Pathology. For a
detailed protocol, see Supplementary Materials.
Statistical evaluation
For statistical analysis, optimal cut-offs to group the cohort
depending on IDO1 protein expression levels (high vs. low) were
determined using the online tool “cut-off Finder” (https://molpathoh
eidelberg.shinyapps.io/CutoffFinder_v1, [31]). Cut-offs were consid-
ered optimal as the point with the most signicant split that was
calculated by a log-rank test. Survival analyses were performed using
IBM SPSS Statistics (Version 27.0.0.0 64-Bit-Version). OS and PFS of the
patient cohort were analyzed with the Kaplan-Meier method and the
Kaplan-Meier estimate of potential follow-up (“reverse Kaplan-Meier")
was used for the calculation of the median follow-up [32]. PFS was
assessed as previously dened [33]. Cox proportional hazard models
were used to further investigate the inuence of IDO1 expression on the
cohort’s survival in univariate and multivariate testing situations. The
additionally considered variables in multiple regressions included the
established clinical parameters age (≤or >60 years), FIGO stage (FIGO
I-II or FIGO III-IV) and the occurrence of residual tumor burden (no
residual or residual tumor). Patients were censored at the time of their
last follow-up in case of a missing dened event for their survival, or
when they were not in follow-up care for PFS at the time of our analyses.
Additionally, possible correlations between IDO1 and the expression of
TIL markers were investigated using data sets generated in previous
studies [33,34]. Due to the exploratory approach of our study, we
refrained from adjustment for multiple testing. P-values of <0.05 were
considered statistically signicant.
Results
IDO1 staining pattern in HGSOC
Immunohistochemical staining revealed mainly cytoplasmic IDO1
protein expression in 500 out of 507 samples (98.62%) with strong and
clear staining quality. Representative images of a stained and annotated
TMA core are shown in Fig. 1A-B. Using the QuPath software, positive
staining was detected at different intensities (weakly, moderately, and
strongly positive, as indicated by different colors in Fig. 1C). In all
subsequent analyses, all IDO1-expressing cells, regardless of staining
intensity, were considered IDO1-positive.
We observed a right-skewed distribution of IDO1-positive cells with
a median proportion of 0.8070% of all cells (IQR: 0.2036 - 3.3805%,
Fig. 1D) and in 492 out of 507 patients, IDO1 expression was detected in
tumor cells. Cut-off determination using a web-based cut-off nder tool
resulted in two different cut-offs for our statistical analyses: for IDO1-
positive tumor cells and OS (cut-off: 2.304) and for IDO1-positive
tumor cells and PFS, respectively (cut-off: 3.2). To eliminate type 2 er-
rors and enhance the overall robustness, we maintained a minimum of
10% for the number of signicant tests for the chosen cut-offs, while the
range of signicant tests varied from 11% to 87.6%, predominantly
exceeding 30%. Hereafter, we will refer to the different groups as either
“IDO1-low” (percentage of IDO1-positive cells equal to or below the
optimal cut-off) or “IDO1-high” (percentage of IDO1-positive cells above
the optimal cut-off).
High IDO1 protein expression in tumor cells correlates with better OS and
PFS
The median OS in the IDO1-low group (n =342) was 38.9 months
(95% CI: 34.0-43.9 months, Supplementary Table S4) and signicantly
lower than the median OS of 59.7 months (95% CI: 40.4-79.3 months) in
the IDO1-high group (n == 165, p <0.001, Fig. 2A). Univariate Cox
regression revealed that an elevated IDO1 expression alone had a sig-
nicant positive effect on OS (HR=0.622, 95% CI: 0.481-0.805, p <
0.001). Even after the inclusion of other relevant risk factors, such as age
(≤or >60 years), FIGO stage (FIGO I-II or FIGO III-IV) and residual
tumor burden, in a multivariate Cox proportional hazard model, the
protein expression level of IDO1 remained an independent signicant
Fig. 1. Images of TMA core in QuPath work-
ow and percentage of IDO1-positive cells in
tumor tissue. (A) Representative close-up image
section of an IDO1-stained TMA core without
annotations. (B) Image of the same TMA core
section including manual annotations of tumor
(red) and stroma areas (green). (C) TMA core
section after positive cell detection with the
color code: green =stroma cells, blue =nega-
tive tumor cells, yellow =weakly positive
tumor cells, orange =moderately positive
tumor cells and red =strongly positive tumor
cells. (D) Distribution of the fraction of IDO1-
positive tumor cells. (For interpretation of the
references to color in this gure legend, the
reader is referred to the web version of this
article.)
I. Hoffmann et al.
Neoplasia 44 (2023) 100934
4
prognostic factor (HR=0.696, 95% CI: 0.518-0.935, p == 0.016), while
among the other covariates, only residual tumor burden also had a
signicant impact (p <0.001).
Similar results were found for PFS. Patients in the IDO1-high group
(n == 92) showed a signicantly prolonged PFS (p == 0.016, Fig. 2B)
with a median survival time of 25.0 months (95% CI: 15.7-34.4 months,
Table. 2), compared to those in the IDO1-low group (n == 241) that had
a median PFS of 18.9 months (95% CI: 16.0-21.8 months). Cox regres-
sion analyses revealed a signicantly improved survival on a univariate
level (HR=0.709, 95% CI: 0.535-0.939, p == 0.016), as well as an
almost signicant impact in combination with the standard prognostic
covariates age, FIGO stage and residual tumor burden (HR=0.736, 95%
CI: 0.535-1.011, p == 0.059).
Higher expression of IDO1 mRNA is linked to a better OS and PFS
To validate our ndings in an independent cohort, we analyzed the
impact of IDO1 mRNA expression on survival using the web-based
Kaplan-Meier plotter tool. We found that a higher expression of IDO1
mRNA signicantly correlated with improved OS (median survival in
months: IDO1-low=38.77 and IDO1-high=50.00, p <0.001, Fig. 3A), as
well as with improved PFS (median survival in months: IDO1-
low=16.00, IDO1-high=19.02, p == 0.003, Fig. 3B).
Increased IDO1 expression correlates with an increased number of TILs
Since IDO1 is known to affect T cell immunity [35,36], we investi-
gated possible correlations of IDO1 with different T cell subsets. Data on
protein expression of the T cell markers CD3, CD4 and CD8 were
available for 119, 105 and 101 patients, respectively [34]. The number
of IDO1-positive tumor cells signicantly correlated with the number of
CD3-positive cells (n == 119, Spearman’s
ρ
=0.609, p <0.001;
Fig. 4A+D), CD4-positive cells (n == 105, Spearman’s
ρ
=0.419, p <
0.001, Fig. 4B+D), and CD8-positive cells (n == 101, Spearman’s
ρ
=0.582, p <0.001, Fig. 4C+D), respectively.
Additionally, we clustered the cohort based on the respective
expression levels and compared the OS and PFS between the different
groups. For each survival analysis, four groups were created based on
Fig. 2. Kaplan-Meier survival curves of overall and progression-free survival with survival tables for IDO1-low versus IDO1-high. (A) Overall survival, grouped by
IDO1 protein expression in tumor cells. (B) Progression-freeIDO1 protein expression in tumor cells and progression-free survival.
Fig. 3. Kaplan Meier plot of survival rates for IDO1 mRNA expression in publicly available data sets of HGSOC patients. (A) Overall survival: n =1144, p =
0.000063, HR=0.74 (0.62-0.88), median survival in months: IDO1-low=38.77 and IDO1-high=50.00. (B) Progression-free survival: n =1029, p =0.003, HR=0.78
(0.66-0.92), median survival in months: IDO1-low=16.00, IDO1-high=19.02. N =number of cases, HR =hazard ratio.
I. Hoffmann et al.
Neoplasia 44 (2023) 100934
5
their respective cut-offs determined with the Cutoff-Finder: IDO1-high/
TILs-high, IDO1-high/TILs-low, IDO1-low/TILs-high and IDO1-low/
TILs-low. We found a signicantly prolonged OS and PFS in both the
groups IDO1-high/CD3-high (OS: p <0.001, PFS: p <0.001,
Fig. S1A+D) and IDO1-high/CD8-high (OS: p <0.001, PFS: p == 0.002,
Fig. S1C +F). For IDO1/CD4+TILs, the best survival rates were found in
the subgroup IDO1-high/CD4-low (Fig. S1B+E).
IDO1 gene expression correlates with enhanced antigen presentation on
tumor cells and is linked to a pro-inammatory tumor microenvironment
To analyze potential mechanisms which are correlated with IDO1
gene expression and might contribute to differences in patient survival,
we performed gene set enrichment analyses in the TCGA ovarian cancer
dataset using the Hallmark and KEGG gene sets [27,28]. Within the
Hallmark gene sets, the gene signatures “interferon alpha response” and
“interferon gamma response” showed the strongest correlation with
IDO1 gene expression (Fig. 5A+B). Furthermore, we found many other
inammation-related gene signatures positively correlated with IDO1
gene expression, such as “allograft rejection”, “IL6/JAK/STAT3
signaling”, “inammatory response” and “TNFalpha signaling via
NFkappaB”. Within the KEGG gene sets, correspondingly, IDO1 gene
expression was positively correlated with multiple inammation-related
gene signatures, such as “antigen processing and presentation”
(Fig. 5C+D). In contrast, in both the Hallmark and the KEGG gene sets,
we observed a negative correlation with “TGFbeta signaling” gene sig-
natures. These results indicate that IDO1 gene expression might be
correlated with a pro-inammatory tumor microenvironment (TME).
In the TCGA bulk gene expression data, the expression of genes or
gene signatures cannot be attributed to distinct cell populations.
Therefore, we analyzed two public single-cell gene expression datasets
of ovarian cancer to study the expression of IDO1 and correlation with
gene signatures on the single-cell level (Fig. 6A). In both datasets, IDO1
was expressed mainly in tumor cells and myeloid dendritic cells
(Fig. 6B). Interestingly, IDO1 gene expression in tumor cells was
heterogeneous across patients (Fig. 6C). Within the myeloid immune
cells, IDO1 was specically expressed in LAMP3+/CLEC9A+dendritic
cells (Fig. 6D). IDO1-positive tumor cells were characterized by higher
expression scores of the Hallmark gene signatures “interferon gamma
response” and “interferon alpha response” and lower expression scores
of the “TGFbeta signaling” signature (Fig. 6E). Additionally, IDO1-pos-
itive tumor cells exhibited higher expression scores of the KEGG gene
signature “antigen processing and presentation” (Fig. 7E), which was
substantiated by other antigen processing and presentation-related gene
signatures as well as individual genes, such as CD74, B2M and different
HLA genes (Fig. S2).
Together, our results indicate that high IDO1 gene expression in
tumor cells is linked to tumor cell response to interferon alpha and
gamma as well as higher capability of antigen processing and presen-
tation by tumor cells which might sustain a pro-inammatory TME.
Interferon-gamma increases IDO1 in stimulated HGSOC cell lines
To validate the association between IFNG response and increased
IDO1 expression, we selected three ovarian cancer cell lines: two cell
lines reported in the literature as possibly HGSOC with TP53 mutations
(OVCAR-3 and SKOV-3) and one cell line dened as unlikely HGSOC and
without TP53 mutations ([37,38], Supplementary Fig. S3A). The cells
were stimulated with 1000U/ml interferon-gamma for 24 hours and
subsequently proteins were harvested for western blotting (Supple-
mentary Fig. S3B). As expected, interferon-gamma increased the protein
expression of IDO1 compared to unstimulated controls and the effect
was more pronounced in OVCAR-3 (possibly HGSOC, TP53_Mut) and
SKOV-3 (possibly HGSOC, TP53Mut) compared to OAW-42 an unlikely
HGSOC cell line with TP53_WT status (Supplementary Fig. S3C+D).
Discussion
In this retrospective study, we demonstrate that IDO1 protein and
mRNA expression serve as a positive prognostic marker for both OS and
Fig. 4. Association of IDO1 with tumor-inltrating lymphocytes. (A) Correlation of number of IDO1-positive cells with number CD3-positive cells. (B) Correlation of
number of IDO1-positive cells with number of CD4-positive cells. (C) Correlation of number of IDO1-positive cells with number of CD8-positive cells. (D) Repre-
sentative HGSOC TMA core sections with immunohistochemical stainings IDO1 (top left), CD3 (bottom left), CD4 (top right) and CD8 (bottom right).
I. Hoffmann et al.
Neoplasia 44 (2023) 100934
6
PFS in a large independent HGSOC cohorts. Furthermore, we show that
high IDO1 protein expression correlates with increased numbers of
CD3+, CD4+and CD8+TILs, and high IDO1 mRNA expression is
associated with an enhanced response to pro-inammatory cytokines,
especially IFNG, providing a potential mechanistic link to improved
patient survival.
To date, IDO1 expression has been widely studied in various cancer
types. While some studies report a negative correlation between IDO1
expression and prognosis, such as in endometrial cancer [39], hormone
receptor-positive breast cancer [40], non-small cell lung cancer [41],
colorectal cancer [42], prostate cancer [43], and glioblastoma [44],
there are also studies showing a positive correlation between IDO1
expression and prognosis, such as in hepatocellular carcinoma [9,45],
basal-like breast carcinoma [46], and rectal cancer [47] . To our
knowledge, three studies by Okamoto et al. [12], Takao et al. [13], and
Inaba et al. [14] specically investigated IDO1 protein expression in
serous ovarian cancer and found an association with poor patient sur-
vival. However, these studies are notably limited by small cohort sizes of
24, 33 and 60 patients, while in our cohort data on OS and PFS was
available for 507 and 333 patients, respectively. Moreover, all three
previous studies applied a semi-quantitative assessment of IDO1 IHC,
whereas we applied an automated quantitative image analysis, allowing
a more precise cut-off determination and reducing observer bias. In
another study, Feng et al. [11] analyzed gene expression data from
TCGA in different gynecologic and breast cancers, including ovarian
serous cystadenocarcinoma, and found that high IDO1 mRNA expression
correlates with improved survival, which is in line with the ndings of
our study.
IDO1 is a well-known target of IFNG signaling [5]. Consistent with
our study, it has previously been shown that high IDO1 mRNA
Fig. 5. Correlation of gene signatures with IDO1 expression in the TCGA HGSOC dataset. (A) Gene set analysis (GSEA) enrichment scores of Hallmark gene sets,
genes ranked by Spearman correlation coefcient with IDO1 expression, only signicant enrichment scores shown (Bonferroni-adjusted p < 0.05). (B) GSEA plot for
Hallmark gene sets indicated in (A). (C) GSEA enrichment scores of KEGG gene sets, genes ranked by Spearman correlation coefcient with IDO1 expression, only
signicant enrichment scores shown (Bonferroni-adjusted p < 0.05). (D) GSEA plot for KEGG gene sets indicated in (C).
I. Hoffmann et al.
Neoplasia 44 (2023) 100934
7
expression correlates with immune-related hallmarks, including IFNG
and IFN-alpha (IFNA) response, in several gynecologic cancers,
including ovarian serous cystadenocarcinoma [11]. Since interferons act
via auto- and paracrine signaling in the TME, different cell types can
respond with increased IDO1 expression. Just as in our study, IDO1
protein expression was associated with improved survival in renal cell
carcinoma [48]. Here, IDO1 was mainly expressed by endothelial cells
and the authors concluded that tumor growth might be restricted, and
survival improved by limiting the inux of tryptophan from the blood to
the tumor cells. Ishio et al. [45] identied IDO1 as a necessary factor for
the antitumor immune response of tumor-inltrating cells and found it
to be expressed only in TILs, but not tumor cells, leading to the
hypothesis that IDO1 expression by TILs might lead to a TME that is
depleted of tryptophan, suppressing tumor proliferation. They also
correlated IDO1 mRNA expression in tumorous tissues with the
expression of IFNG and TNF-alpha mRNA, raising the possibility that
IDO1 is expressed due to the presence of these cytokines, which might be
produced by activated TILs. In our HGSOC cohort, we found IDO1 to be
mainly expressed in tumor cells. This was underlined by single-cell gene
expression data of two independent HGSOC cohorts showing IDO1
mRNA expression mainly in tumor cells and a subset of dendritic cells,
but not in TILs. In cell culture experiments, we found that IDO1 protein
expression was inducible by IFNG stimulation in two possibly HGSOC
cell lines, but not in the unlikely HGSOC cell line, supporting the
Fig. 6. Gene expression of IDO1 on single-cell level in two ovarian cancer datasets. (A) Visualization of two ovarian cancer single-cell datasets by uniform manifold
approximation and projection (UMAP), color-coded by cell type. (B) Gene expression of IDO1 in UMAP plots as shown in (A). (C) Proportion of IDO1-positive and
IDO1-negative tumor epithelial cells across different patient samples. (D) Gene expression of IDO1 in different immune cell subtypes. (E) Module scores of indicated
Hallmark and KEGG gene sets in IDO1-positive and IDO1-negative tumor epithelial cells.
I. Hoffmann et al.
Neoplasia 44 (2023) 100934
8
hypothesis that IDO1 expression in tumor cells might be induced by
IFNG signaling in HGSOC.
The interaction between IDO1 and TILs is complex. IDO1 expression
within tumors can lead to the depletion of tryptophan and the accu-
mulation of kynurenine in the TME. This metabolic shift can create an
immunosuppressive environment that inhibits the activity of TILs and
promotes tumor immune evasion. On the other hand, TILs can produce
IFNG, which can induce the expression of IDO1 in tumor cells and other
immune cells. This creates a feedback loop where IDO1 expression by
the tumor cells further suppresses TILs, leading to a dampened immune
response against the cancer. In line with that, several studies have found
a correlation between an increased expression of IDO1 and a reduced
number of CD8+cells in ovarian cancer [14,49], endometrial cancer [7]
and esophageal squamous cell carcinoma [8]. Other studies detected a
correlation with an additional reduction of CD4+cells [50] or a reduced
number of CD3+cells [42]. While a few studies could not detect any
signicant coherences between IDO1 and TILs [40], our analyses
showed a signicant association of elevated IDO1 levels with an
increased number of such, corresponding to the ndings of Li et al. [9] in
hepatocellular carcinoma, Toda et al. [10] in osteosarcoma and Feng
et al. [11] in several gynecologic and breast cancers. The controversial
ndings regarding correlation of IDO1 and TILs in different tumor types
point to potentially context-dependent mechanisms regulating IDO1
expression and TIL inltration.
Our study’s ndings suggest that the role of IDO1 in ovarian cancer is
more complex than originally anticipated. Specically, IDO1 appears to
be benecial for patients by mediating the suppression of tumor growth.
Induced by IFNG, which is produced by natural killer cells, natural killer
T cells, CD4, Th1, and CD8 cytotoxic T lymphocyte effector T cells as
part of the innate and antigen-specic immunity, it is conceivable that
the positive effect of IDO1 is caused by a pro-inammatory TME, leading
to higher IFNG levels and, thus, increased IDO1 expression. Feng et al.
(2020) provided additional explanations for the positive effect of IDO1,
suggesting that the interaction between tumor cells and TILs expressing
IDO1 inuences and changes the TME, leading to different outcomes in
different cancers. Furthermore, IDO1 can deprive tumor cells of tryp-
tophan, leading to decreased proliferation. Taken together, it remains
elusive if IDO1 itself mechanistically contributes to improved survival or
serves as a surrogate biomarker for a high anti-tumor immune response
mediated by IFNG signaling and TIL inltration.
At rst glance, it may seem debatable whether the very low and
specic cut-offs we determined for IDO1 expression are feasible in a
clinical setting. However, this alleged aw can be easily overcome by
employing digital methods for the analysis of immunohistochemical
staining, which can provide more accurate and reproducible results.
Regarding the TILs subgroups and single-cell analyses we performed, it
should be noted that the available datasets included only a relatively
small number of patients (TILs maximum n == 119, single-cell: n ==
19), allowing for explorative analyses only. Nevertheless, our ndings
provide valuable insights into the expression patterns of IDO1 in indi-
vidual cells and suggest potential mechanisms underlying the observed
associations between IDO1 expression and immune cell inltration.
Further studies with larger patient cohorts are needed to conrm and
extend our ndings.
Our study clearly demonstrates that IDO1 is a positive prognostic
marker in HGSOC. We identied a correlation between IDO1 expression
and the immune response, specically a positive correlation with
increased numbers of CD3+, CD4+and CD8+TILs. We propose that
further studies are needed to clarify the exact role of IDO1 in HGSOC,
especially in the light of modern aspects of immune modulating therapy
concepts.
CRediT authorship contribution statement
Inga Hoffmann: Conceptualization, Data curation, Formal analysis,
Investigation, Methodology, Writing – original draft, Writing – review &
editing. Mihnea P. Dragomir: Data curation, Formal analysis, Funding
acquisition, Investigation, Methodology, Writing – original draft,
Writing – review & editing. Nanna Monj´
e: Data curation, Formal
analysis, Writing – review & editing. Carlotta Keunecke: Writing –
review & editing. Catarina Alisa Kunze: Writing – review & editing.
Simon Schallenberg: Writing – review & editing. Sofya Marchenko:
Writing – review & editing. Wolfgang D. Schmitt: Writing – review &
editing. Hagen Kulbe: Writing – review & editing. Jalid Sehouli:
Writing – review & editing. Ioana Elena Braicu: Writing – review &
editing. Paul Jank: Writing – review & editing. Carsten Denkert:
Writing – review & editing. Silvia Darb-Esfahani: Writing – review &
editing. David Horst: Writing – review & editing. Bruno V. Sinn:
Writing – review & editing. Christine Sers: Supervision, Writing – re-
view & editing. Philip Bischoff: Conceptualization, Data curation,
Formal analysis, Funding acquisition, Investigation, Methodology,
Writing – original draft, Writing – review & editing. Eliane T. Taube:
Conceptualization, Formal analysis, Investigation, Methodology, Su-
pervision, Writing – original draft, Writing – review & editing.
Declaration of Competing Interest
The authors declare that they have no known competing nancial
interests or personal relationships that could have appeared to inuence
the work reported in this paper.
Acknowledgements
We would like to thank Ines Koch and Sylwia Handzik for their
outstanding technical assistance.
Funding
M.P.D. is a participant in the BIH-Charit´
e Junior Clinical Scientist
Program funded by the Charit´
e–Universit¨
atsmedizin Berlin and the
Berlin Institute of Health. M.P.D.’s work is funded by a DKTK Berlin
Young Investigator Grant 2022 and Berliner Krebsgesellschaft
(DRFF202204). PB is participant in the BIH-Charit´
e Clinician Scientist
Program funded by the Charit´
e-Universit¨
atsmedizin Berlin and the
Berlin Institute of Health. PB’s work is funded by a DKTK Berlin Young
Investor Grant 2022.
Supplementary materials
Supplementary material associated with this article can be found, in
the online version, at doi:10.1016/j.neo.2023.100934.
References
[1] Sung, H.A.-O., et al., Global Cancer Statistics 2020: GLOBOCAN Estimates of
Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. 2021(1542-
4863 (Electronic)).
[2] B.M. Reid, J.B. Permuth, T.A. Sellers, Epidemiology of ovarian cancer: a review,
Cancer Biol. Med. 14 (1) (2017) 9–32.
[3] L.A. Torre, et al., Ovarian cancer statistics, CA Cancer J. Clin. 68 (4) (2018)
284–296, 2018.
[4] J. Zhu, L. Yan, Q. Wang, Efcacy of PD-1/PD-L1 inhibitors in ovarian cancer: a
single-arm meta-analysis, J. Ovarian Res. 14 (1) (2021) 112.
[5] R. Yoshida, et al., Induction of pulmonary indoleamine 2,3-dioxygenase by
interferon, Proc. Natl. Acad. Sci. U S A 78 (1) (1981) 129–132.
[6] I. Theate, et al., Extensive proling of the expression of the indoleamine 2,3-
dioxygenase 1 protein in normal and tumoral human tissues, Cancer Immunol. Res.
3 (2) (2015) 161–172.
[7] K. Ino, et al., Inverse correlation between tumoral indoleamine 2,3-dioxygenase
expression and tumor-inltrating lymphocytes in endometrial cancer: its
association with disease progression and survival, Clin. Cancer Res. 14 (8) (2008)
2310–2317.
[8] G. Zhang, et al., Involvement of indoleamine 2,3-dioxygenase in impairing tumor-
inltrating CD8 T-cell functions in esophageal squamous cell carcinoma, Clin. Dev.
Immunol. (2011), 384726, 2011.
[9] S. Li, et al., Mechanism and prognostic value of indoleamine 2,3-dioxygenase 1
expressed in hepatocellular carcinoma, Cancer Sci. 109 (12) (2018) 3726–3736.
I. Hoffmann et al.
Neoplasia 44 (2023) 100934
9
[10] Y. Toda, et al., PD-L1 and IDO1 expression and tumor-inltrating lymphocytes in
osteosarcoma patients: comparative study of primary and metastatic lesions,
J. Cancer Res. Clin. Oncol. 146 (10) (2020) 2607–2620.
[11] X. Feng, et al., A comprehensive analysis of IDO1 expression with tumour-
inltrating immune cells and mutation burden in gynaecologic and breast cancers,
J. Cell. Mol. Med. 24 (9) (2020) 5238–5248.
[12] A. Okamoto, et al., Indoleamine 2,3-dioxygenase serves as a marker of poor
prognosis in gene expression proles of serous ovarian cancer cells, Clin. Cancer
Res. 11 (16) (2005) 6030–6039.
[13] M. Takao, et al., Increased synthesis of indoleamine-2,3-dioxygenase protein is
positively associated with impaired survival in patients with serous-type, but not
with other types of, ovarian cancer, Oncol. Rep. 17 (6) (2007) 1333–1339.
[14] T. Inaba, et al., Role of the immunosuppressive enzyme indoleamine 2,3-dioxyge-
nase in the progression of ovarian carcinoma, Gynecol. Oncol. 115 (2) (2009)
185–192.
[15] F. Li, et al., IDO1: An important immunotherapy target in cancer treatment, Int.
Immunopharmacol. 47 (2017) 70–77.
[16] M. Platten, R. Offringa, Cancer immunotherapy: exploiting neoepitopes, Cell Res.
25 (8) (2015) 887–888.
[17] W.C.o. Tumours, Female Genital Tumours, 5th ed., T.W.C.o.T.E. Board, 2020. Vol.
4.
[18] P. Bankhead, et al., QuPath: Open source software for digital pathology image
analysis, Sci. Rep. 7 (1) (2017) 16878.
[19] Gyorffy, B., Z. L´
anczky A Fau - Sz´
all´
asi, and Z. Sz´
all´
asi, Implementing an online
tool for genome-wide validation of survival-associated biomarkers in ovarian-
cancer using microarray data from 1287 patients. 2012 (1479-6821 (Electronic)).
[20] B. Gyorffy, Discovery and ranking of the most robust prognostic biomarkers in
serous ovarian cancer, Geroscience (2023).
[21] E. Cerami, et al., The cBio cancer genomics portal: an open platform for exploring
multidimensional cancer genomics data, Cancer Discov. 2 (5) (2012) 401–404.
[22] J. Gao, et al., Integrative analysis of complex cancer genomics and clinical proles
using the cBioPortal, Sci. Signal 6 (269) (2013) pl1.
[23] T. Wu, et al., clusterProler 4.0: A universal enrichment tool for interpreting omics
data, Innovation 2 (3) (2021), 100141.
[24] S. Olalekan, et al., Characterizing the tumor microenvironment of metastatic
ovarian cancer by single-cell transcriptomics, Cell Rep. 35 (8) (2021), 109165.
[25] Olbrecht, S.A.-O., et al., High-grade serous tubo-ovarian cancer rened with single-
cell RNA sequencing: specic cell subtypes inuence survival and determine
molecular subtype classication. 2021 (1756-994X (Electronic)).
[26] Y. Hao, et al., Integrated analysis of multimodal single-cell data, Cell 184 (13)
(2021) 3573–3587, e29.
[27] A. Liberzon, et al., The Molecular Signatures Database (MSigDB) hallmark gene set
collection, Cell Syst. 1 (6) (2015) 417–425.
[28] M. Kanehisa, et al., KEGG for taxonomy-based analysis of pathways and genomes,
Nucleic. Acids. Res. 51 (D1) (2023) D587–D592.
[29] Thompson, J.A.-O., et al., Gene signature of antigen processing and presentation
machinery predicts response to checkpoint blockade in non-small cell lung cancer
(NSCLC) and melanoma. LID - 10.1136/jitc-2020-000974 [doi] LID - e000974.
2020 (2051-1426 (Electronic)).
[30] R. Chen, et al., Antigen presentation machinery signature-derived CALR mediates
migration, polarization of macrophages in glioma and predicts immunotherapy
response, Front. Immunol. 13 (2022), 833792.
[31] Budczies, J., et al., Cutoff Finder: a comprehensive and straightforward Web
application enabling rapid biomarker cutoff optimization. 2012 (1932-6203
(Electronic)).
[32] M. Schemper, T.L. Smith, A note on quantifying follow-up in studies of failure time,
Controlled Clinic. Trials 17 (4) (1996) 343–346.
[33] S. Marchenko, et al., Prognostic value of regulatory T cells and T helper 17 cells in
high grade serous ovarian carcinoma, J. Cancer Res. Clin. Oncol. (2022).
[34] Darb-Esfahani, S., et al., Prognostic impact of programmed cell death-1 (PD-1) and
PD-ligand 1 (PD-L1) expression in cancer cells and tumor-inltrating lymphocytes
in ovarian high grade serous carcinoma. 2015 (1949-2553 (Electronic)).
[35] A. Mellor, Indoleamine 2,3 dioxygenase and regulation of T cell immunity,
Biochem. Biophys. Res. Commun. 338 (1) (2005) 20–24.
[36] L. Zhai, et al., IDO1 in cancer: a Gemini of immune checkpoints, Cell Mol.
Immunol. 15 (5) (2018) 447–457.
[37] E. Papp, et al., Integrated genomic, epigenomic, and expression analyses of ovarian
cancer cell lines, Cell Rep. 25 (9) (2018) 2617–2633.
[38] A. Bairoch, The cellosaurus, a cell-line knowledge resource, J. Biomol. Tech. 29 (2)
(2018) 25–38.
[39] K. Ino, et al., Indoleamine 2,3-dioxygenase is a novel prognostic indicator for
endometrial cancer, Br. J. Cancer 95 (11) (2006) 1555–1561.
[40] D.E. Carvajal-Hausdorf, et al., Objective measurement and clinical signicance of
IDO1 protein in hormone receptor-positive breast cancer, J. Immunother. Cancer 5
(1) (2017) 81.
[41] S. Astigiano, et al., Eosinophil granulocytes account for indoleamine 2,3-dioxyge-
nase-mediated immune escape in human non-small cell lung cancer, Neoplasia 7
(4) (2005) 390–396.
[42] G. Brandacher, et al., Prognostic value of indoleamine 2,3-dioxygenase expression
in colorectal cancer: effect on tumor-inltrating T cells, Clin. Cancer Res. 12 (4)
(2006) 1144–1151.
[43] Feder-Mengus, C., et al., High expression of indoleamine 2,3-dioxygenase gene in
prostate cancer. 2008 (1879-0852 (Electronic)).
[44] L. Zhai, et al., Inltrating T cells increase IDO1 expression in glioblastoma and
contribute to decreased patient survival, Clin. Cancer Res. 23 (21) (2017)
6650–6660.
[45] T. Ishio, et al., Immunoactivative role of indoleamine 2,3-dioxygenase in human
hepatocellular carcinoma, J. Gastroenterol. Hepatol. 19 (3) (2004) 319–326.
[46] J. Jacquemier, et al., High expression of indoleamine 2,3-dioxygenase in the
tumour is associated with medullary features and favourable outcome in basal-like
breast carcinoma, Int. J. Cancer 130 (1) (2012) 96–104.
[47] J. Schollbach, et al., Prognostic value of tumour-inltrating CD8+lymphocytes in
rectal cancer after neoadjuvant chemoradiation: is indoleamine-2,3-dioxygenase
(IDO1) a friend or foe? Cancer Immunol. Immunother. 68 (4) (2019) 563–575.
[48] R. Riesenberg, et al., Expression of indoleamine 2,3-dioxygenase in tumor
endothelial cells correlates with long-term survival of patients with renal cell
carcinoma, Clin. Cancer Res. 13 (23) (2007) 6993–7002.
[49] A. Amobi-McCloud, et al., IDO1 expression in ovarian cancer induces PD-1 in T
cells via aryl hydrocarbon receptor activation, Front. Immunol. 12 (2021), 678999.
[50] F. Qian, et al., Efcacy of levo-1-methyl tryptophan and dextro-1-methyl
tryptophan in reversing indoleamine-2,3-dioxygenase-mediated arrest of T-cell
proliferation in human epithelial ovarian cancer, Cancer Res. 69 (13) (2009)
5498–5504.
I. Hoffmann et al.