Serum high-mobility group box 1 as a predictive marker for cytotoxic chemotherapy-induced lung injury in patients with lung cancer and interstitial lung disease

概要

Serum high-mobility group box 1 as a predictive marker for cytotoxic
chemotherapy-induced lung injury in patients with lung cancer and interstitial
lung disease

Satoshi Nakaoa, Kakuhiro Yamaguchia, Hiroshi Iwamotoa, Shinjiro Sakamotoa, Yasushi
Horimasua, Takeshi Masudaa, Shintaro Miyamotoa, Taku Nakashimaa, Shinichiro
Ohshimob, Kazunori Fujitakaa, Hironobu Hamadac, Noboru Hattoria

a

Department of Molecular and Internal Medicine, Graduate School of Biomedical and

Health Sciences, Hiroshima University, Hiroshima, Japan
b

Department of Emergency and Critical Care Medicine, Graduate School of Biomedical

and Health Sciences, Hiroshima University, Hiroshima, Japan
c

Department of Physical Analysis and Therapeutic Sciences, Graduate School of

Biomedical and Health Sciences, Hiroshima University, Hiroshima, Japan

1

Highlights
x

HMGB1 is an inflammatory protein associated with lung cancer and lung injury

x

Lung injury by chemotherapy is common in lung cancer with interstitial lung
disease

x

Elevated HMGB1 levels may be the first blood marker to predict this adverse
event

x

Interstitial lung disease may be a risk factor of the disease by increasing
HMGB1

x

Tumor burden also contributes to disease development by increasing HMGB1

2

Abstract
Background: High-mobility group box 1 (HMGB1) is a pro-inflammatory protein, that
is associated with tumorigenesis, interstitial lung disease (ILD), and acute lung injury.
Chemotherapy-induced lung injury is a common and serious adverse event in patients
with lung cancer and ILD, but its pathogenesis and predictive biomarkers are not known.
This study aimed to investigate the predictive potential of serum HMGB1 levels for
cytotoxic chemotherapy-induced lung injury in these patients.

Methods: From 743 patients with advanced lung cancer, we enrolled 83 consecutive
patients with ILD and background-matched 83 patients without ILD. Additionally, 83
healthy subjects were included. After measuring baseline levels of serum HMGB1 in
three groups, we evaluated the predictive values of baseline HMGB1 levels for cytotoxic
chemotherapy-induced lung injury in patients with lung cancer and ILD.

Results: Higher levels of serum HMGB1 were independently associated with higher
tumor burden, as assessed by total tumor size, and the presence of ILD. Twenty-five
(30.1%) of patients with lung cancer and ILD experienced cytotoxic chemotherapyinduced lung injury within one year. Univariate Cox proportional hazards model

3

showed that higher levels of HMGB1 and higher tumor burden were associated with
disease onset. Moreover, multivariate analysis revealed that only HMGB1 was
independently associated with this severe complication in patients with lung cancer and
ILD.

Conclusions: HMGB1 is a potential predictive blood biomarker for cytotoxic
chemotherapy-induced lung injury in patients with lung cancer and ILD. This study also
suggests a potential pathogenesis of this serious adverse event that tumor- and ILDderived HMGB1 accelerates lung injury.

Keywords: lung cancer, interstitial lung disease, lung injury, high-mobility group box 1,
soluble receptor for advanced glycation end products, biomarker

Abbreviations: ALAT, Latin American Thoracic Association; ATS, American Thoracic
Society; AUC, area under the curve; CT, computed tomography; ERS, European
Respiratory Society; FVC, forced vital capacity; HMGB1, high-mobility group box 1;
ILD, interstitial lung disease; IPF, idiopathic pulmonary fibrosis; IQR, interquartile
range; JRS, Japanese Respiratory Society; PS, performance status; RAGE, receptor for

4

advanced glycation end products; RECIST, Response Evaluation Criteria in Solid
Tumors; ROC, receiver operating characteristic; sRAGE, soluble receptor for advanced
glycation end products; UIP, usual interstitial pneumonia

5

1. Introduction

Interstitial lung disease (ILD) is a risk factor of lung cancer, and ILD is
observed in 7.5% to 15.2% of patients with lung cancer at diagnosis [1, 2]. Treatment
options are limited in patients with lung cancer and ILD due to the high incidence of
treatment-related exacerbation of ILD [3]. Cytotoxic chemotherapy is usually selected
as the treatment of choice in these patients with advanced lung cancer, but cytotoxic
chemotherapy-induced lung injury has been observed in 13.3–34.9% of patients with
lung cancer and ILD [4-9]; furthermore, the pathogenesis and predictive blood markers
of this potentially fatal adverse event remain unclear.

High-mobility group box 1 (HMGB1) is one of the damage-associated
molecular patterns released by injured cells. HMGB1 binds to cell surface receptors,
such as the receptor for advanced glycation end product (RAGE), and this interaction
accelerates pro-inflammatory intracellular signaling associated with tumorigenesis and
acute lung injury, which includes acute exacerbation of ILD [10, 11]. It was shown that
HMGB1 was highly expressed in lung cancer tissue [12], and circulatory HMGB1
levels were shown to be increased by larger tumor size and decreased by surgical
resection of the tumor [13]. Additionally, a previous report showed that the presence of
ILD elevated HMGB1 circulatory levels, and importantly, these higher levels at the

6

baseline could predict earlier onset of acute exacerbation of ILD [11]. Based on these
observations, we hypothesized that the levels of circulatory HMGB1 would be
increased in patients with lung cancer and ILD, and HMGB1 could be a predictive
blood biomarker for cytotoxic chemotherapy-induced lung injury in these patients.
To elucidate this hypothesis, first, we measured serum HMGB1 levels in the
following three groups: lung cancer patients with ILD, those without ILD, and healthy
controls. Second, we evaluated the association of HMGB1 serum levels with tumor
burden and the presence of ILD among lung cancer patients with and without ILD.
Third, the potential of HMGB1 as a predictive biomarker of cytotoxic chemotherapyinduced lung injury in patients with lung cancer and ILD was analyzed. Finally, because
circulatory soluble RAGE (sRAGE) acts as a decoy receptor for HMGB1 and inhibits
HMGB1-associated inflammation [14-17], we evaluated the association of sRAGE with
HMGB1 and cytotoxic chemotherapy-induced lung injury.

2. Material and Methods
2. 1. Study population and design
As shown in Supplementary Fig. 1, the present study included 83 consecutive
advanced lung cancer patients with ILD and background-matched 83 patients without

7

ILD from 743 patients who were diagnosed and treated at the Hiroshima University
Hospital between October 2003 and December 2018. Eighty-three healthy subjects were
also enrolled. Lung cancer patients without ILD were matched to those with ILD for
age, sex, smoking history, performance status, stage, and histological type. Advanced
lung cancer was defined as unresectable stage III/stage IV/postoperative recurrence of
disease, using the TNM classification [18], and patients who were treated with cytotoxic
chemotherapy after blood draw were included for the present study. In patients with
lung cancer and ILD, this study excluded those who received radical thoracic radiation
therapy and those who were treated with tyrosine kinase inhibitors targeted to oncogenic
driver mutation. This study was approved by the Ethics Committee of Hiroshima
University Hospital (M326) and all of the participants provided written informed
consent.

2. 2. Diagnostic criteria for ILD and cytotoxic chemotherapy-induced lung injury
The diagnostic criteria for ILD were the existence of bilateral reticulation and
consolidation or ground-glass attenuation on pre-treatment computed tomography (CT),
which was performed within one month from starting the treatment. The usual
interstitial pneumonia (UIP) pattern was defined as UIP and probable UIP in the CT

8

pattern of the American Thoracic Society, European Respiratory Society, Japanese
Respiratory Society, and Latin American Thoracic Association (ATS/ERS/JRS/ALAT)
clinical practice guideline of idiopathic pulmonary fibrosis (IPF) [19]. These were
independently classified by two pulmonologists blinded to the patient clinical details.
The concordance rate was 84.3% (70/83). In 13 cases with conflicting results, the third
experienced pulmonologist additionally and independently evaluated the CT images,
and the final decision was made by the majority vote.
Cytotoxic chemotherapy-induced lung injury was diagnosed using the criteria
modified from those for IPF, to suit the practical aspects of cancer chemotherapy [20,
21]. The criteria were as follows:
i.

Acute worsening or development of dyspnea within one month.

ii.

CT showing new ground-glass abnormality bilaterally and/or consolidation.

iii.

No indication of apparent heart failure, pulmonary invasion of lung cancer,
or pulmonary infection; no improvement with antibiotic treatment and
negative sputum and/or blood cultures.

iv.

Development of cytotoxic chemotherapy-induced lung injury within one
month of administration of the last cytotoxic chemotherapy.

9

2. 3. Measurement of serum biomarkers and tumor burden
Serum samples were collected before chemotherapy administration at our
hospital and stored at -80 °C. Serum levels of HMGB1 and sRAGE were measured using
commercially available enzyme-linked immunosorbent assay kits according to
manufacturer’s instructions (HMGB1 ELISA Kit II [Shino-Test Corporation, Tokyo] and
Human RAGE Quantikine ELISA Kit [R&D Systems, Minneapolis, MN, USA],
respectively).
Tumor burden was quantified as the sum of the diameters (longest axis for nonnodal lesions and short axis for nodal lesions) of measurable target lesions using
unidimensional Response Evaluation Criteria in Solid Tumors (RECIST)-defined
measurements. Measurable target lesions (≥10 mm in the longest diameter for nonnodal lesions and ≥15 mm in the short axis for nodal lesions) were selected on baseline
CT scans, allowing up to two lesions per organ and up to five lesions in total [22].

2. 4. Statistical analysis
Values were expressed as medians (interquartile ranges [IQRs]). Differences
among groups were evaluated using Pearson's chi-squared and Kruskal–Wallis tests. If
there was a significant difference in the Kruskal–Wallis test for multiple comparisons,

10

the Mann–Whitney U test with Bonferroni correction was performed for individual
comparisons. Spearman’s correlation coefficient was performed to investigate the
correlation of HMGB1 serum levels with tumor burden and sRAGE serum levels, and
linear regression analysis was performed in order to find out factors that affect serum
levels of HMGB1. Receiver operating characteristic (ROC) curve analysis was
performed to define the optimal cut-off levels of serum HMGB1, tumor burden, and
serum sRAGE for predicting the development of cytotoxic chemotherapy-induced lung
injury within one year of starting cytotoxic chemotherapy. Disease development was
evaluated using Kaplan–Meier analysis and log-rank test. Death due to cancer
progression was considered a censoring event. Cox proportional hazards analysis was
used to identify significant predictors for the development of the disease. P <0.05 were
considered to indicate statistical significance. All data analyses were performed using
JMP, version 14.1.0 (SAS Institute Inc., Cary, NC, USA).

3. Results
3. 1. Cohort characteristics
Although there were missing data, a significant difference only in the baseline
forced vital capacity (FVC) was observed in 74 of 83 lung cancer patients with ILD and

11

61 of 83 lung cancer patients without. Healthy subjects were significantly younger and
had smoked significantly less than lung cancer patients with or without ILD, and had
significantly higher FVC than lung cancer patients with ILD (Table 1). Of the 83 lung
cancer patients with ILD, 77 (92.8%) had idiopathic interstitial pneumonia, 5 had
collagen vascular disease (3, rheumatoid arthritis; 1, systemic sclerosis; 1, mixed
connective tissue disease), and the other one patient had asbestosis. Seven patients were
receiving immunosuppressive treatment (4, only steroid therapy; 3, combination steroid
therapy and immunosuppressant), two were receiving anti-fibrotic drugs, and three
received home oxygen therapy before and during the cancer treatment. Additionally,
two patients received palliative irradiation for bronchial stenosis due to lung cancer
within one year of starting cytotoxic chemotherapy. However, no patient developed lung
injury during the observation period.

3. 2. Baseline serum concentrations of HMGB1
Serum levels of HMGB1 in lung cancer patients with ILD were significantly
higher than those in patients without ILD (5.34 ng/mL [IQR, 2.79–8.46] and 3.77
ng/mL [IQR, 2.48–5.87], P=0.008), and these levels were significantly higher than
those among the healthy controls (2.24 ng/mL [IQR, 1.61–3.55], P<0.001 each) (Fig.

12

1). Additionally, among lung cancer patients with ILD and those without, Spearman’s
correlation coefficient revealed that baseline levels of serum HMGB1 were significantly
and positively correlated with tumor burden (rs=0.438, P<0.001) (n=166) (Fig. 2).
Univariate linear regression analysis showed that older age, higher pack-year smoking
history, the presence of ILD, and higher tumor burden were significantly associated
with higher levels of serum HMGB1. Multivariate linear regression analysis revealed
that only the presence of ILD and higher tumor burden were significantly and
independently associated with higher levels of serum HMGB1 when adjusted by
confounders (Table 2).

3. 3. Predictive potential of HMGB1 and tumor burden for cytotoxic
chemotherapy-induced lung injury
Of the 83 patients with lung cancer and ILD, 25 (30.1%) had cytotoxic
chemotherapy-induced lung injury during the observation period. The frequency and
severity of the different regimens and treatment lines are summarized in Supplementary
Table 1. ROC curve analysis revealed that the optimal cut-off levels for predicting
cytotoxic chemotherapy-induced lung injury were 5.04 ng/mL for serum HMGB1 (area
under the curve [AUC]=0.68, specificity=55.2%, sensitivity=80.0%) and 70.7 mm for

13

tumor burden (AUC=0.61, specificity=53.5%, sensitivity=80.0%) (Supplementary Fig.
2). Kaplan–Meier analysis revealed that not only patients with HMGB1 higher levels
but also those with higher tumor burden had a significantly earlier onset of cytotoxic
chemotherapy-induced lung injury (Fig. 3a, b). Univariate Cox proportional hazards
model revealed that serum HMGB1 and tumor burden higher than 70.7 mm were
significant predictors of the disease onset. Multivariate Cox proportional hazards model
revealed that only serum HMGB1 was the independent predictor when adjusted by
confounders, including tumor burden (Table 3).

3. 4. Correlation of sRAGE with the risk of chemotherapy-induced lung injury
stratified by HMGB1 levels
Among lung cancer patients with ILD and those without ILD, the median
serum level of sRAGE was 751.3 pg/mL (IQR, 505.5-1064.7), which was significantly
and negatively correlated with that of HMGB1 (n=166) (rs=-0.153, P=0.049)
(Supplementary Fig. 3). In patients with lung cancer and ILD, Kaplan–Meier analysis
revealed that patients with sRAGE levels >606.9 pg/mL, which was obtained by ROC
curve analysis (AUC=0.66, specificity=60.3%, sensitivity=76.0%) (Supplementary Fig.
2), had a significantly lower development rate of cytotoxic chemotherapy-induced lung

14

injury (Supplementary Fig. 4); that was particularly noticeable in patients with HMGB1
levels ≥5.04 ng/mL (n=46), but not in those with HMGB1 levels <5.04 ng/mL (n=37)
(Fig. 3c, d).

4. Discussion
Chemotherapy-induced lung injury is a common and important adverse event
in patients with lung cancer and ILD, and this study was conducted to investigate the
first predictive blood biomarker for this life-threatening condition based on the
hypothesis shown in the graphical abstract (Figure 4). We found an independent
association between serum HMGB1 levels and the onset of chemotherapy-induced lung
injury in patients with lung cancer and ILD. Additionally, exploratory analysis revealed
that higher levels of serum sRAGE, a decoy receptor for RAGE ligands, including
HMGB1, was associated with decreased incidence of cytotoxic chemotherapy-induced
lung injury in patients with higher HMGB1 levels. These results indicate that HMGB1
is a promising biomarker and may have a role in the pathogenesis of cytotoxic
chemotherapy-induced lung injury in patients with lung cancer and ILD.
This study found that higher tumor burden was correlated with higher levels of
serum HMGB1; furthermore, these two factors were associated with earlier onset of

15

cytotoxic chemotherapy-induced lung injury in the univariate analysis. Notably, serum
HMGB1 was shown as an independent predictor of chemotherapy-induced lung injury
after adjusting for potential confounders, including tumor burden. These data indicate
that the circulatory levels of HMGB1 play a promising role in the development of
cytotoxic chemotherapy-induced lung injury in patients with lung cancer and ILD, and
higher tumor burden may accelerate this severe complication via increasing circulatory
HMGB1 (Figure 4).
This study also demonstrated that the presence of ILD accompanied with lung
cancer was associated with higher levels of HMGB1 in circulation, which helps to explain
the association between the presence of ILD and the risk of cytotoxic chemotherapyinduced lung injury [4-9]. We and others have previously demonstrated that
increased levels of serum and bronchoalveolar lavage fluid HMGB1 were observed in
patients with ILDs compared to those in healthy subjects, and these higher levels in serum
were associated with the earlier development of acute exacerbation of ILD [11, 23]. A
previous report has shown that increased levels of serum HMGB1 in patients with ILD
may reflect the increased HMGB1 expression in the nuclei of infiltrating inflammatory
cells, alveolar macrophages, and injured epithelial cells [23]. These data indicate that the
presence of ILD could also promote the augmented pro-inflammatory condition in the

16

lungs of lung cancer patients by increasing HMGB1 levels (Figure 4).
We also evaluated serum levels of sRAGE in this study, because we previously
found that increased serum levels of sRAGE were associated with a reduced risk of
acute exacerbation in patients with IPF [17]. As a result, sRAGE higher levels were
associated with reduced incidence of cytotoxic chemotherapy-induced lung injury in
patients with higher levels of HMGB1. These data suggest that the effect of heightened
HMGB1 could be canceled by high levels of sRAGE and support the hypothesis that
HMGB1 is involved in the pathogenesis of chemotherapy-induced lung injury in
patients with lung cancer and ILD. Further investigation is needed to clarify the
possibility of blocking HMGB1 as a therapeutic target.
This study has limitations. First, the present study was conducted in a single
facility, and the sample size was relatively small. Second, non-target lesion by the
RECIST criteria, e.g., pleural dissemination, could not be quantified as regards tumor
burden. Third, the treatment regimen was not the same when patients developed
chemotherapy-induced lung injury because there was no standard of care for patients with
lung cancer and ILD; this is due to the lack of a randomized study evaluating the
differences in the risk of lung injury for each cytotoxic drug. Recently, Haruna et al.
showed that blood HMGB1 levels increased with Docetaxel treatment [24]. This

17

observation suggests that treatment with cytotoxic chemotherapy may further increase
HMGB1 levels (Figure 4). However, it is not clear whether each cytotoxic drug might
have similar effects. Finally, the mechanism for the pathogenesis of drug-induced lung
injury could generally be classified into two, the allergic mechanism and the cytotoxic
mechanism. However, it is practically difficult to do this because no consensus diagnostic
method exists. A prospective study is necessary to confirm the results of this study and
evaluate serial changes in HMGB1 after chemotherapy to elucidate the mechanism of the
association between HMGB1 and cytotoxic chemotherapy-induced lung injury.

5. Conclusions
Increased HMGB1 levels in the circulation were independently associated with
earlier development of cytotoxic chemotherapy-induced lung injury in patients with
lung cancer and ILD, while higher tumor burden and the presence of ILD were
associated with heightened HMGB1. Our results suggest that serum HMGB1 is a
promising biomarker and a potential therapeutic target for this serious adverse event in
patients with lung cancer and ILD.

18

References
[1] T. Omori, M. Tajiri, T. Baba, T. Ogura, T. Iwasawa, K. Okudela, T. Takemura, M.S.
Oba, T. Maehara, H. Nakayama, M. Tsuboi, M. Masuda, Pulmonary Resection for Lung
Cancer in Patients With Idiopathic Interstitial Pneumonia, Ann Thorac Surg 100(3)
(2015) 954-60.
[2] H. Kawasaki, K. Nagai, T. Yokose, J. Yoshida, M. Nishimura, K. Takahashi, K.
Suzuki, R. Kakinuma, Y. Nishiwaki, Clinicopathological characteristics of surgically
resected lung cancer associated with idiopathic pulmonary fibrosis, J Surg Oncol 76(1)
(2001) 53-7.
[3] Y. Minegishi, K. Takenaka, H. Mizutani, J. Sudoh, R. Noro, T. Okano, A. Azuma, A.
Yoshimura, M. Ando, E. Tsuboi, S. Kudoh, A. Gemma, Exacerbation of idiopathic
interstitial pneumonias associated with lung cancer therapy, Intern Med 48(9) (2009)
665-72.
[4] H. Kenmotsu, T. Naito, M. Kimura, A. Ono, T. Shukuya, Y. Nakamura, A. Tsuya, K.
Kaira, H. Murakami, T. Takahashi, M. Endo, N. Yamamoto, The risk of cytotoxic
chemotherapy-related exacerbation of interstitial lung disease with lung cancer, J
Thorac Oncol 6(7) (2011) 1242-6.
[5] S. Niho, K. Goto, K. Yoh, Y.H. Kim, H. Ohmatsu, K. Kubota, N. Saijo, Y.

19

Nishiwaki, Interstitial shadow on chest CT is associated with the onset of interstitial
lung disease caused by chemotherapeutic drugs, Jpn J Clin Oncol 36(5) (2006) 269-73.
[6] K. Isobe, Y. Hata, S. Sakamoto, Y. Takai, K. Shibuya, S. Homma, Clinical
characteristics of acute respiratory deterioration in pulmonary fibrosis associated with
lung cancer following anti-cancer therapy, Respirology (Carlton, Vic.) 15(1) (2010) 8892.
[7] P. Camus, S. Kudoh, M. Ebina, Interstitial lung disease associated with drug therapy,
Br J Cancer 91 Suppl 2 (2004) S18-23.
[8] S. Kudoh, H. Kato, Y. Nishiwaki, M. Fukuoka, K. Nakata, Y. Ichinose, M. Tsuboi, S.
Yokota, K. Nakagawa, M. Suga, H. Jiang, Y. Itoh, A. Armour, C. Watkins, T.
Higenbottam, F. Nyberg, Interstitial lung disease in Japanese patients with lung cancer:
a cohort and nested case-control study, Am J Respir Crit Care Med 177(12) (2008)
1348-57.
[9] Y. Minegishi, N. Kokuho, Y. Miura, M. Matsumoto, A. Miyanaga, R. Noro, Y. Saito,
M. Seike, K. Kubota, A. Azuma, K. Kida, A. Gemma, Clinical features, anti-cancer
treatments and outcomes of lung cancer patients with combined pulmonary fibrosis and
emphysema, Lung cancer (Amsterdam, Netherlands) 85(2) (2014) 258-63.
[10] H. Shimizu, S. Sakamoto, T. Isshiki, K. Furuya, A. Kurosaki, S. Homma,

20

Association of serum high-mobility group box protein 1 level with outcomes of acute
exacerbation of idiopathic pulmonary fibrosis and fibrosing nonspecific interstitial
pneumonia, PLoS One 13(5) (2018) e0196558.
[11] K. Yamaguchi, H. Iwamoto, S. Sakamoto, Y. Horimasu, T. Masuda, S. Miyamoto,
T. Nakashima, S. Ohshimo, K. Fujitaka, H. Hamada, N. Hattori, Serum high-mobility
group box 1 is associated with the onset and severity of acute exacerbation of idiopathic
pulmonary fibrosis, Respirology 25 (2020) 275-280.
[12] K.K. Sun, C. Ji, X. Li, L. Zhang, J. Deng, N. Zhong, X.Y. Wu, Overexpression of
high mobility group protein B1 correlates with the proliferation and metastasis of lung
adenocarcinoma cells, Mol Med Rep 7(5) (2013) 1678-82.
[13] G.H. Shang, C.Q. Jia, H. Tian, W. Xiao, Y. Li, A.H. Wang, L. Dong, D.J. Lin,
Serum high mobility group box protein 1 as a clinical marker for non-small cell lung
cancer, Respir Med 103(12) (2009) 1949-53.
[14] H. Yonekura, Y. Yamamoto, S. Sakurai, R.G. Petrova, M.J. Abedin, H. Li, K. Yasui,
M. Takeuchi, Z. Makita, S. Takasawa, H. Okamoto, T. Watanabe, H. Yamamoto, Novel
splice variants of the receptor for advanced glycation end-products expressed in human
vascular endothelial cells and pericytes, and their putative roles in diabetes-induced
vascular injury, Biochem J 370(Pt 3) (2003) 1097-109.

21

[15] I.H. Park, S.I. Yeon, J.H. Youn, J.E. Choi, N. Sasaki, I.H. Choi, J.S. Shin,
Expression of a novel secreted splice variant of the receptor for advanced glycation end
products (RAGE) in human brain astrocytes and peripheral blood mononuclear cells,
Mol Immunol 40(16) (2004) 1203-11.
[16] J. Xie, J.D. Mendez, V. Mendez-Valenzuela, M.M. Aguilar-Hernandez, Cellular
signalling of the receptor for advanced glycation end products (RAGE), Cell Signal
25(11) (2013) 2185-97.
[17] K. Yamaguchi, H. Iwamoto, Y. Horimasu, S. Ohshimo, K. Fujitaka, H. Hamada, W.
Mazur, N. Kohno, N. Hattori, AGER gene polymorphisms and soluble receptor for
advanced glycation end product in patients with idiopathic pulmonary fibrosis,
Respirology 22(5) (2017) 965-971.
[18] P. Goldstraw, K. Chansky, J. Crowley, R. Rami-Porta, H. Asamura, W.E. Eberhardt,
A.G. Nicholson, P. Groome, A. Mitchell, V. Bolejack, The IASLC Lung Cancer Staging
Project: Proposals for Revision of the TNM Stage Groupings in the Forthcoming
(Eighth) Edition of the TNM Classification for Lung Cancer, Journal of thoracic
oncology : official publication of the International Association for the Study of Lung
Cancer 11(1) (2016) 39-51.
[19] G. Raghu, M. Remy-Jardin, J.L. Myers, L. Richeldi, C.J. Ryerson, D.J. Lederer, J.

22

Behr, V. Cottin, S.K. Danoff, F. Morell, K.R. Flaherty, A. Wells, F.J. Martinez, A.
Azuma, T.J. Bice, D. Bouros, K.K. Brown, H.R. Collard, A. Duggal, L. Galvin, Y.
Inoue, R.G. Jenkins, T. Johkoh, E.A. Kazerooni, M. Kitaichi, S.L. Knight, G. Mansour,
A.G. Nicholson, S.N.J. Pipavath, I. Buendía-Roldán, M. Selman, W.D. Travis, S. Walsh,
K.C. Wilson, Diagnosis of Idiopathic Pulmonary Fibrosis. An Official
ATS/ERS/JRS/ALAT Clinical Practice Guideline, Am J Respir Crit Care Med 198(5)
(2018) e44-e68.
[20] S. Nakao, K. Yamaguchi, S. Sakamoto, Y. Horimasu, T. Masuda, S. Miyamoto, T.
Nakashima, H. Iwamoto, K. Fujitaka, H. Hamada, N. Hattori, Chemotherapy-associated
Acute Exacerbation of Interstitial Lung Disease Shortens Survival Especially in Small
Cell Lung Cancer, Anticancer Res 39(10) (2019) 5725-5731.
[21] H.R. Collard, C.J. Ryerson, T.J. Corte, G. Jenkins, Y. Kondoh, D.J. Lederer, J.S.
Lee, T.M. Maher, A.U. Wells, K.M. Antoniou, J. Behr, K.K. Brown, V. Cottin, K.R.
Flaherty, J. Fukuoka, D.M. Hansell, T. Johkoh, N. Kaminski, D.S. Kim, M. Kolb, D.A.
Lynch, J.L. Myers, G. Raghu, L. Richeldi, H. Taniguchi, F.J. Martinez, Acute
Exacerbation of Idiopathic Pulmonary Fibrosis. An International Working Group
Report, Am J Respir Crit Care Med 194(3) (2016) 265-75.
[22] E.A. Eisenhauer, P. Therasse, J. Bogaerts, L.H. Schwartz, D. Sargent, R. Ford, J.

23

Dancey, S. Arbuck, S. Gwyther, M. Mooney, L. Rubinstein, L. Shankar, L. Dodd, R.
Kaplan, D. Lacombe, J. Verweij, New response evaluation criteria in solid tumours:
revised RECIST guideline (version 1.1), Eur J Cancer 45(2) (2009) 228-47.
[23] N. Hamada, T. Maeyama, T. Kawaguchi, M. Yoshimi, J. Fukumoto, M. Yamada, S.
Yamada, K. Kuwano, Y. Nakanishi, The role of high mobility group box1 in pulmonary
fibrosis, American journal of respiratory cell and molecular biology 39(4) (2008) 440-7.
[24] M. Haruna, M. Hirata, K. Iwahori, T. Kanazawa, Y. Yamamoto, K. Goto, A.
Kawashima, A. Morimoto-Okazawa, S. Funaki, Y. Shintani, A. Kumanogoh, H. Wada,
Docetaxel Upregulates HMGB1 Levels in Non-small Cell Lung Cancer, Biol Pharm
Bull 43(3) (2020) 399-403.

24

Table 1. Demographic data of the study cohort
Lung cancer

Lung cancer

with ILD

without ILD

Subjects, n

83

83

83

Age, years

71 (67–78)

71 (63–75)

52 (50–55) ***

Male

74

69

71

Female

9

14

12

Smoking history, pack-years

50.0 (32.0–80.0)

46.0 (29.8–75.0)

10.0 (0.0–30.0) ***

FVC, % predicted̋

87.4 (70.6-100.0)

94.3 (78.6-106.5) ###

95.6 (85.4-107.4) ###

0–1

66

72

-

≥2

17

11

-

III

19

10

-

IV

59

68

-

Recurrence

5

5

-

Adenocarcinoma

34

47

-

Squamous cell carcinoma

15

12

-

Small cell carcinoma

29

23

-

Others

5

1

-

 UIP pattern

24

-

-

 Non-UIP pattern

59

-

-



Control

Sex

PS

Stage

Histological type

ILD pattern

* P<0.05, ** P<0.01, and *** P<0.001 compared to lung cancer patients with ILD
and those without ILD, and # P<0.05, ## P<0.01, and ### P<0.001 compared to lung
cancer patients with ILD, Mann–Whitney U test with Bonferroni correction for
individual comparisons
̋

FVC was measured only in 74 of 83 lung cancer patients with ILD and 61 of 83

25

lung cancer patients without.
Data are presented as median and IQR.
Abbreviations: FVC, forced vital capacity; ILD, interstitial lung disease; IQR,
interquartile range; PS, performance status; UIP, usual interstitial pneumonia

26

Table 2. Correlations between serum levels of HMGB1 and baseline characteristics in
lung cancer patients and healthy subjects (n=249)
Variables

β

T

P-value

Age, years

0.28

4.58

<0.001***

Sex, male

-0.10

-1.56

0.119

Smoking history, pack-years

0.26

4.15

<0.001***

0.29

4.74

<0.001***

0.47

8.32

<0.001***

Age, years

0.01

0.09

0.932

Sex, male

-0.06

-1.08

0.281

Smoking history, pack-years

-0.01

-0.01

0.995

0.14

2.22

0.028*

0.41

5.78

<0.001***

Univariate analysis

 Presence of ILD
Tumor burden, mm
Multivariate analysis

 Presence of ILD
Tumor burden, mm

* P<0.05, ** P<0.01, and *** P<0.001 linear regression analysis
Abbreviations: HMGB1, high-mobility group box 1; ILD, interstitial lung disease

27

Table 3. Cox proportional hazards model for predicting the onset of cytotoxic
chemotherapy-induced lung injury in patients with lung cancer and ILD (n=83)
Variables

HR

95% CI

P-value

Age, years

0.97

0.93–1.01

0.115

Sex, male

1.79

0.53–11.15

0.392

PS ≥2

1.79

0.51–4.93

0.331

Smoking history, pack-years

0.99

0.98–1.00

0.184

2.19

0.95-4.84

0.065

Tumor burden, ≥70.7 mm

3.91

1.58-11.78

0.002**

Tumor burden, mm

1.01

0.99-1.02

0.070

HMGB1, ng/mL

1.13

1.05–1.20

0.001**

sRAGE, pg/mL

0.99

0.996–0.999

0.002**

1.65

0.68-3.87

0.264

Tumor burden, ≥70.7 mm

2.55

0.98-7.91

0.056

HMGB1, ng/mL

1.09

1.02-1.16

0.020*

sRAGE, pg/mL

0.99

0.99-1.00

0.074

Univariate analysis

 UIP pattern

Multivariate analysis
 UIP pattern

* P<0.05, ** P<0.01, and *** P<0.001 Cox proportional hazards model
Abbreviations: CI, confidence interval; HMGB1, high-mobility group box 1; HR,
hazard ratio; ILD, interstitial lung disease; PS, performance status; sRAGE, soluble
receptor for advanced glycation end products; UIP, usual interstitial pneumonia

28

Figure legends
Figure 1. Baseline levels of serum HMGB1
In the order of lung cancer patients with interstitial lung disease (ILD), those
without ILD, and healthy subjects, serum levels of high-mobility group box 1
(HMGB1) were significantly decreased. Boxes represent the 25th to 75th percentiles;
solid lines within the boxes show the median values; whiskers represent the 10th and
90th percentiles; the circles represent outliers. * P<0.05, ** P<0.01, and *** P<0.001
using the Mann–Whitney U test with Bonferroni correction. IQR, interquartile range

Figure 2. Correlation between serum HMGB1 levels and tumor burden
In lung cancer patients with and without interstitial lung disease, serum levels of
high-mobility group box 1 (HMGB1) were significantly and positively correlated with
tumor burden (n=166) (rs=0.438 P<0.001, using Spearman’s correlation coefficient).

Figure 3. Kaplan–Meier analysis of the onset of cytotoxic chemotherapy-induced lung
injury in patients with lung cancer and ILD
Kaplan-Meier analysis revealed that patients with higher serum levels of highmobility group box 1 (HMGB1) (a) and higher tumor burden (b) had a shorter follow-

29

up period before the development of cytotoxic chemotherapy-induced lung injury.
Soluble receptor for advance glycation end product (sRAGE) acts as a HMGB1
inhibitor, and its serum levels higher than 606.9 pg/mL were significantly decreased the
development rate of this severe complication in patients with HMGB1 levels ≥5.04
ng/mL (c) but not in those with HMGB1 levels <5.04 ng/mL (d) (using log-rank test).
ILD, interstitial lung disease

Figure 4. Schematic diagram of the proposed molecular mechanism of HMGB1induced pro-inflammatory condition, induced by lung cancer, ILD, and cytotoxic
chemotherapy
Up-regulation of high-mobility group box 1 (HMGB1) by the presence of lung
cancer and interstitial lung disease (ILD) induce pro-inflammatory conditions via
intracellular signaling from the receptor for advanced glycation end products (RAGE).
Soluble RAGE (sRAGE) exerts an anti-inflammatory effect, suppressing proinflammatory intracellular signaling by neutralizing HMGB1 as a decoy receptor.

30

Figure 1

31

Figure 2

32

Figure 3

33

34

Figure 4.

35

Supplementary Table 1. Chemotherapy regimens and treatment lines for cytotoxic
chemotherapy-induced lung injury within one year of starting chemotherapy
No. of patients with

No. of patients without

Grade of lung injury*,

lung injury, n

lung injury, n

1–2/3–4/5

First line
Platinum agents+ETP

3

27

1/1/1

Platinum agents+PEM

1

2

0/0/1

Platinum agents+S-1

0

2

0/0/0

Platinum agents+GEM

0

2

0/0/0

CBDCA+PTX

4

13

1/0/3

CBDCA+nab-PTX

0

4

0/0/0

CDDP+DTX

0

2

0/0/0

S-1

2

3

0/1/1

VNR

2

7

0/1/1

DTX

0

4

0/0/0

PEM

1

3

1/0/0

GEM

0

1

0/0/0

Platinum agents+ETP

0

2

0/0/0

CBDCA+PTX

0

2

0/0/0

CDDP+PEM+BEV

0

1

0/0/0

CDDP+VNR

0

1

0/0/0

S-1

1

4

0/0/1

DTX

3

6

2/1/0

PEM

1

2

0/1/0

AMR

3

7

0/1/2

nab-PTX

0

2

0/0/0

NGT

1

1

0/1/0

CPT-11

0

1

0/0/0

S-1

0

3

0/0/0

VNR

1

1

0/0/1

Second line

Third line

36

DTX

1

1

0/1/0

PEM

1

1

0/1/0

GEM

0

2

0/0/0

AMR

0

1

0/0/0

nab-PTX

0

1

0/0/0

PTX

0

1

0/0/0

CPT-11

0

1

0/0/0

Nivolumab

0

1

0/0/0

S-1

0

1

0/0/0

VNR

0

1

0/0/0

Atezolizumab

0

1

0/0/0

0

1

0/0/0

Fourth line

Fifth line
PEM

*According to the Common Terminology Criteria for Adverse Events version 4.0.
Abbreviation: AMR, amrubicin; BEV, bevacizumab; CBDCA, carboplatin; CDDP, cisplatin; CPT-11,
irinotecan; DTX, docetaxel; ETP, etoposide; GEM, gemcitabine; NGT, nogitecan; PEM, pemetrexed;
PTX, paclitaxel; S-1, Tegafur/Gimeracil/Oteracil; VNR, vinorelbine

37

Supplementary Fig. 1. Flowchart of patient enrollment

Among 743 patients with lung cancer with prospectively collected blood samples, 83
consecutive patients with lung cancer and interstitial lung disease (ILD) were enrolled.
Additionally, background-matched 83 lung cancer patients without ILD and 83 healthy controls
were also enrolled.

38

Supplementary Fig. 2. Receiver operating characteristic curve analysis for predicting the onset
of cytotoxic chemotherapy-induced lung injury in patients with lung cancer and ILD (n=83)

Receiver operating characteristic curve analysis revealed that the optimal cut-off levels
for predicting cytotoxic chemotherapy-induced lung injury were 5.04 ng/mL for serum highmobility group box 1 (HMGB1) (area under the curve [AUC]=0.68, specificity=55.2%,
sensitivity=80.0%), 70.7 mm for tumor burden (AUC=0.61, specificity=53.5%,
sensitivity=80.0%), and 606.9 pg/mL for soluble receptor for advanced glycation end products
(sRAGE) (AUC=0.66, specificity=60.3%, sensitivity=80.0%). CI, confidence interval; ILD,
interstitial lung disease

39

Supplementary Fig. 3. Correlation between serum HMGB1 and sRAGE levels

In lung cancer patients with and without interstitial lung disease (n=166), serum levels
of high-mobility group box 1 (HMGB1) were significantly and negatively correlated with those
of soluble receptor for advanced glycation end products (sRAGE) (rs=-0.153, P=0.049, using
Spearman’s correlation coefficient).

40

Supplementary Fig. 4. Kaplan–Meier analysis of the onset of cytotoxic chemotherapy-induced
lung injury in patients with lung cancer and ILD

Higher levels of serum soluble receptor for advanced glycation end products (sRAGE)
were associated with a lower development rate of cytotoxic chemotherapy-induced lung injury
(n=83) (using log-rank test). ILD, interstitial lung disease

41