Cancer Immunology, Immunotherapy

https://doi.org/10.1007/s00262-020-02677-7

ORIGINAL ARTICLE

Human leukocyte antigen expression in paired primary lung tumors

and brain metastases in non‑small cell lung cancer
Jarrett J. Failing1 · Marie Christine Aubry2 · Aaron S. Mansfield1 

Received: 13 May 2020 / Accepted: 17 July 2020

© Springer-Verlag GmbH Germany, part of Springer Nature 2020

Abstract
Loss of human leukocyte antigen (HLA) class 1 expression is a mechanism of tumor immune escape and may contribute to
resistance to immunotherapy. Patients with non-small cell lung cancer (NSCLC) treated with immune checkpoint inhibitors
can have discordant responses between brain metastases and extracranial sites of disease. We sought to evaluate whether
HLA class 1 expression was retained in metastatic NSCLC. Patients with paired primary NSCLC and brain metastases were
identified from our institution’s tissue registry. HLA class 1 cell membrane expression on tumor cells was determined by
immunohistochemistry. Tumors with greater than the median of 10% HLA expression were considered positive. Agreement
statistics (κ) were used to assess the congruence of HLA expression. 51 patients with paired primary NSCLC and brain
lesions were identified. The median HLA class 1 expression was 20% in the primary tumors (IQR 0–65%) and 10% in the
brain metastases (IQR 5–40%). 27 primary tumors and 24 brain metastases were positive for HLA expression. There was
disagreement in HLA positivity between paired lesions in 11 patients (22%, 95% CI 12–35%) (κ = 0.57, 95% CI 0.35–0.79)
(p = 0.0001). None of the patients received checkpoint inhibitors for treatment of these lesions. The results show that while
there is moderate agreement in HLA class 1 expression between primary lung tumor and brain metastasis pairs, HLA expres-
sion is incongruent in nearly one quarter of patients. Loss of antigen presentation may represent one of the many potential
mechanisms of discordant responses to checkpoint inhibitor therapy.

Keywords  Non-small cell lung cancer · Brain metastases · Immunotherapy · Human leukocyte antigen (HLA) class 1

Abbreviations Introduction
ICI Immune checkpoint inhibitor
TIL Tumor-infiltrating lymphocyte Immune checkpoint inhibitor (ICI) therapy has drastically
TME Tumor microenvironment changed the treatment and prognosis of patients with meta-
static non-small cell lung cancer (NSCLC) in recent years.
Brain metastases develop in up to 30–40% of patients with
NSCLC [1] and are associated with an unfavorable prog-
nosis [2]. Patients with untreated brain metastases were
excluded in many of the pivotal clinical trials of ICIs in
NSCLC. Our understanding of the effects of ICIs on brain
metastases comes from smaller clinical trials and retro-
Electronic supplementary material  The online version of this spective studies—which have found that brain metastases
article (https​://doi.org/10.1007/s0026​2-020-02677​-7) contains respond to ICI therapy in up to 30% of patients. However,
supplementary material, which is available to authorized users. 13–21% of patients show discordant responses between the
brain metastases and extracranial sites of disease [3, 4].
* Aaron S. Mansfield
mansfield.aaron@mayo.edu Understanding the reasons for these discrepant responses
could help improve the outcomes for these patients.
1
Division of Medical Oncology, Department of Oncology, The tumor microenvironment (TME) has emerged as a
Mayo Clinic, 200 First Street SW, Rochester, MN 55905, significant factor in disease progression and the response
USA
to ICIs [5]. We have previously shown that there can be
2
Department of Laboratory Medicine and Pathology, Mayo significant differences in the TME of paired primary lung
Clinic, Rochester, MN, USA

13
Vol.:(0123456789)

cancers and brain metastases in individual patients [6, 7].
The extent of tumor-infiltrating lymphocytes (TILs) and
expression of PD-L1 on tumor and immune cells can be
discordant between the brain metastasis and primary tumor
[6]. In addition, there are fewer T cell clones in brain metas-
tases compared to paired primary lung lesions despite a
high tumor mutation burden in the brain metastases [7]. It is
unclear whether these differences are due to decreased traf-
ficking of T cells through the blood–brain barrier, decreased
antigen presentation, or other methods of immune escape
by the cancer. Human leukocyte antigen (HLA) class 1 is
crucial for cytotoxic T cell responses to cancer, and loss of
HLA class 1 expression is a mechanism of tumor immune
evasion and may contribute to resistance to ICIs [8–10]. In
this project, we sought to evaluate the agreement of HLA
class 1 expression between paired primary lung lesions and
brain metastases in NSCLC.
Materials and methods
Adult patients with both a primary NSCLC tumor and brain
metastasis specimen were identified from Mayo Clinic’s
Tissue Registry. Specimens were collected per institu-
tional protocols, and the study was approved by the Mayo
Clinic’s Institutional Review Board (IRB #13-007990). A
pathologist (MCA) reviewed the tissue sections for pres-
ence of tumor and adequacy. Tissue sectioning and IHC
staining was performed using the Leica Bond RX stainer
(Leica Biosystems Inc, Buffalo Grove, IL). Formalin fixed
paraffin embedded -tissue blocks were sectioned at 5 μm
and IHC staining was performed on-line. Slides for the
HLA-ABC stain were retrieved for 20 min using Epitope
Retrieval one (Leica Biosystems Inc, Buffalo Grove, IL) and
incubated in Dako Protein Block (Aglient, Santa Clara, CA)
for 5 min. An HLA-ABC primary antibody (Ms Monoclo-
nal clone:EMR8-5, Abcam ab70328, Cambridge, MA) was
diluted to 1:4000 in Dako Background Reducing Diluent
(Aglient, Santa Clara, CA) and incubated for 15 min. The
detection system used was Polymer Refine Detection Sys-
tem (Leica Biosystems Inc, Buffalo Grove, IL). This system
includes the hydrogen peroxidase block, post primary and
polymer reagent, DAB, and Hematoxylin. Immunostaining
visualization was achieved by incubating slides 10 min in
DAB and DAB buffer (1:19 mixture) from the Bond Polymer
Refine Detection System. To this point, slides were rinsed
between steps with 1X Bond Wash Buffer (Leica Biosys-
tems Inc, Buffalo Grove, IL). Slides were counterstained
for 5 min using Schmidt hematoxylin and molecular biology
grade water (1:1 mixture), followed by several rinses in 1X
Bond wash buffer and distilled water. Once completed, slides
were removed from the stainer and rinsed in tap water for
5 min. Slides were dehydrated in increasing concentrations
13
Cancer Immunology, Immunotherapy
of ethyl alcohol and cleared in three changes of xylene
prior to permanent coverslipping in xylene-based medium.
Membranous HLA expression was recorded as a percent of
tumor cell positivity, in increments of 5%. Since there is no
standard value defining HLA positivity, we used the median
HLA expression of all the specimens analyzed (10%) as a
cutoff for being HLA positive or negative. Data on TILs and
PD-L1 expression were available on many of these cases
from a prior project [6]. Agreement statistics (κ) were used
to assess the congruence of HLA expression. Fisher’s exact
test, paired T test, Wilcoxon rank sum test, Spearman’s rank
correlation (ρ), Kaplan–Meier method and log rank test were
used as appropriate for statistical analysis using JMP ver-
sion 14.1.0. GraphPad Prism version 8.0.1 was used for the
violin plot. P values < 0.05 were considered significant in
the exploratory retrospective analysis.
Results
We analyzed 51 patients with paired primary lung cancer
tumors and brain metastases. The specimens dated from the
years 1994 to 2015. The majority of the patients (92%) had
no extra-cranial sites of metastatic disease at the time of
the brain metastasis resection. The median time between
the primary lung tumor resection and the brain metasta-
sis resection was 13.5 months [inter-quartile range (IQR)
5.8–30.7 months, complete range 0–166 months]. Thirteen
patients (25%) had time between the resections of less than
6 months. Nine patients received chemotherapy, four patients
received radiation to the brain, and no patients received
ICI therapy between the primary lung tumor resection and
brain metastasis resection. Only two patients subsequently
received ICIs (Table 1).
The median membranous HLA class 1 expression was
20% (IQR 0–65%) versus 10% (IQR 5–40%) and the mean
HLA class 1 expression was 32% versus 27% in the primary
lung tumors and brain metastases respectively (p = 0.1149)
(Fig. 1). The tumor cells of 27 primary lung tumors and
24 brain metastases were positive for HLA. There was
congruent HLA expression in 40 cases (78%, 95% CI
65–88%) and incongruent expression in 11 cases (22%,
95% CI 12–35%) (κ = 0.57, 95% CI 0.35–0.79) (p = 0.0001)
(Fig. 2). In seven cases the primary lung tumor was positive
but the brain metastasis was negative, and in four cases the
primary lung tumor was negative and the brain metastasis
was positive. There was no significant difference in the time
between the primary tumor and brain metastasis resections
in patients with congruent HLA expression compared to
those with incongruent HLA expression (12.3 months ver-
sus 13.9 months respectively, p = 0.6634). There was no
significant difference in overall survival from the time of
brain metastasis resection between patients with congruent
Cancer Immunology, Immunotherapy

Table 1  Patient Characteristics positive, congruent negative, or incongruent HLA expres-

Total patients 51 sion (p = 0.3427) (Supplementary Fig. 1). From a previous
study, the PD-L1 expression and extent of CD3 + TILs were
Sex, n (%) known in 46 and 42 of the patients respectively [6]. There
 Female 29 (57%) was no agreement between PD-L1 and HLA class 1 (congru-
 Male 22 (43%) ent in 46% of patient, κ = − 0.108) or between CD3 + TILs
Median age at diagnosis, years (range) 61 (36–80) and HLA class 1 expression (congruent in 50% of patients,
Stage at diagnosis, n (%) κ = − 0.094) (Supplementary Table 1). There was also no
 I 16 (31%) association between PD-L1 and HLA class 1 (ρ = 0.02,
 II 13 (25%) p = 0.8953) or between CD3 + TILs and HLA class 1 expres-
 III 11 (22%) sion as continuous variables (ρ = 0.06, p = 0.7120).
 IV 11 (22%)
No extracranial metastatic disease at time of brain 47 (92%)
metastasis resection, n (%)
Discussion
Histology, n (%)
 Adenocarcinoma 39 (76%)
We found there is moderate agreement in HLA class 1
 Squamous Cell 12 (24%)
expression via immunohistochemistry between paired pri-
Tobacco Use, n (%)
mary lung tumors and brain metastases in patients with
 Current/Former 45 (88%)
non-small cell lung cancer. However, nearly one-quarter of
 Never 6 (12%)
patients have incongruent HLA expression. Our results are
Mutation Status, n (%)
similar to a previous study looking at HLA changes at the
 EGFR, ALK, ROS, KRAS 4 (8%)
genomic level, in which 22% of patients had differences in
 Not detected 9 (18%)
loss of heterozygosity in HLA in the brain metastasis versus
 Not tested 38 (75%)
the lung primary [11]. While we don’t know the response to
Median time between lesions—months (IQR) 13.5 (5.8–30.7)
ICIs in our cohort, other studies have reported a similar rate
 < 6 months, n (%) 13 (25%)
of discordant responses (13–21%) to checkpoint inhibitors
Treatment between lesions, n (%)
between brain metastases and extracranial sites of disease
 Immunotherapy 0 (0%)
[3, 4].
 Chemotherapy 9 (18%)
We found no significant difference in overall survival
 Brain radiation 4 (8%)
from the time of brain metastasis resection between patients
 None 39 (77%)
with congruent positive, congruent negative, or incongru-
Immunotherapy after brain metastasis resection, n 2 (4%)
(%)
ent HLA expression; however, the size of our cohort lim-
its conclusions on this, and only two patients subsequently
received any ICI therapy. A potential limitation to the gen-
eral applicability of our findings to all NSCLC patients
with brain metastases is that the majority of the patients
in the cohort had oligometastatic disease with only CNS
involvement (as these are the patients who typically undergo
resection of brain metastases). Furthermore, a larger cohort
of more contemporaneous paired lesions may further influ-
ence our understanding of heterogeneity of HLA expression
between disparate sites of disease. For these reasons, our
results should be considered hypothesis-generating and the
heterogeneity of HLA will hopefully be profiled in larger
context of the immune landscape.
Loss of antigen presentation has been associated with
resistance to immune checkpoint inhibitors, but it may also
be a component of metastatic heterogeneity with clonal
evolution. Although our study is limited by its retrospec-
tive nature and lack of globally accepted thresholds of HLA
expression by NSCLC, these results suggest that clonal evo-
Fig. 1  Violin plot of the percentage HLA class 1 expression in all of lution or metastatic drivers may affect antigen presentation
the specimens by tumors. Accordingly, loss of antigen presentation may

13

Fig. 2  Representative patients
for congruent positive (a),
congruent negative (b), and
incongruent (c) HLA class 1
expression in the lung primary
tumor and brain metastasis
(Hematoxylin and Eosin stain,
HLA immunostain, ×200)
represent one of the many potential mechanisms of discord-
ant responses to checkpoint inhibitor therapy.
Author contributions  All authors contributed to the study conception
and design. Material preparation and data collection were performed
by Marie Christine Aubry. Data analysis were performed Jarrett Failing
and Aaron Mansfield. The first draft of the manuscript was written by
Jarrett Failing and all authors revised the manuscript. All authors read
and approved the final manuscript.
Funding  Dr. Mansfield was supported by National Institutes of Health
Grant P30CA015083.
4. Goldberg SB, Gettinger SN, Mahajan A, Herbst RS, Chiang
Data availability  The datasets used during the current study are avail-
able from the corresponding author on reasonable request.
Compliance with ethical standards  J Clin Oncol 36(15_suppl):2009–2009. https​://doi.org/10.1200/
Conflict of interest Dr. Mansfield reports research support from
Novartis, and Verily; remuneration to his institution for participation
on advisory boards for AbbVie, BMS, Astra Zeneca and Genentech;
travel support from Roche, and is a nonremunerated director of the
Mesothelioma Applied Research Foundation. Dr. Failing and Dr.
Aubry report no conflicts of interest.
Ethical approval  The study was approved by the Mayo Clinic’s Insti-
tutional Review Board (IRB #13-007990).
References
1. Riihimaki M, Hemminki A, Fallah M, Thomsen H, Sundquist K,
Sundquist J, Hemminki K (2014) Metastatic sites and survival in
13
Cancer Immunology, Immunotherapy
lung cancer. Lung Cancer 86(1):78–84. https​://doi.org/10.1016/j.
lungc​an.2014.07.020
2. Ali A, Goffin JR, Arnold A, Ellis PM (2013) Survival of patients
with non-small-cell lung cancer after a diagnosis of brain metas-
tases. Curr Oncol 20(4):e300–306. https​: //doi.org/10.3747/
co.20.1481
3. Hendriks LEL, Henon C, Auclin E, Mezquita L, Ferrara R, Aud-
igier-Valette C, Mazieres J, Lefebvre C, Rabeau A, Le Moulec S,
Cousin S, Duchemann B, le Pechoux C, Botticella A, Ammari
S, Gazzah A, Caramella C, Adam J, Lechapt E, Planchard D,
De Ruysscher D, Dingemans AM, Besse B (2019) Outcome of
patients with non-small cell lung cancer and brain metastases
treated with checkpoint inhibitors. J Thorac Oncol 14(7):1244–
1254. https​://doi.org/10.1016/j.jtho.2019.02.009
AC, Lilenbaum R, Jilaveanu L, Rowen E, Gerrish H, Komlo
A, Wei W, Chiang V, Kluger HM (2018) Durability of brain
metastasis response and overall survival in patients with non-
small cell lung cancer (NSCLC) treated with pembrolizumab.
JCO.2018.36.15_suppl​.2009
5. Bodor JN, Boumber Y, Borghaei H (2019) Biomarkers for immune
checkpoint inhibition in non-small cell lung cancer (NSCLC).
Cancer. https​://doi.org/10.1002/cncr.32468​
6. Mansfield AS, Aubry MC, Moser JC, Harrington SM, Dronca
RS, Park SS, Dong H (2016) Temporal and spatial discordance
of programmed cell death-ligand 1 expression and lymphocyte
tumor infiltration between paired primary lesions and brain metas-
tases in lung cancer. Ann Oncol 27(10):1953–1958. https​://doi.
org/10.1093/annon​c/mdw28​9
7. Mansfield AS, Ren H, Sutor S, Sarangi V, Nair A, Davila J, Els-
bernd LR, Udell JB, Dronca RS, Park S, Markovic SN, Sun Z,
Halling KC, Nevala WK, Aubry MC, Dong H, Jen J (2018) Con-
traction of T cell richness in lung cancer brain metastases. Sci Rep
8(1):2171. https​://doi.org/10.1038/s4159​8-018-20622​-8
8. Perea F, Sanchez-Palencia A, Gomez-Morales M, Bernal M, Con-
cha A, Garcia MM, Gonzalez-Ramirez AR, Kerick M, Martin J,
Garrido F, Ruiz-Cabello F, Aptsiauri N (2018) HLA class I loss
Cancer Immunology, Immunotherapy
and PD-L1 expression in lung cancer: impact on T-cell infiltra-
tion and immune escape. Oncotarget 9(3):4120–4133. https​://doi.
org/10.18632​/oncot​arget​.23469​
9. Gettinger S, Choi J, Hastings K, Truini A, Datar I, Sowell R,
Wurtz A, Dong W, Cai G, Melnick MA, Du VY, Schlessinger
J, Goldberg SB, Chiang A, Sanmamed MF, Melero I, Agorreta
J, Montuenga LM, Lifton R, Ferrone S, Kavathas P, Rimm DL,
Kaech SM, Schalper K, Herbst RS, Politi K (2017) Impaired
HLA Class I Antigen Processing and Presentation as a Mecha-
nism of Acquired Resistance to Immune Checkpoint Inhibitors
in Lung Cancer. Cancer Discov 7(12):1420–1435. https​://doi.
org/10.1158/2159-8290.CD-17-0593
1 0. Aptsiauri N, Ruiz-Cabello F, Garrido F (2018) The transition from
HLA-I positive to HLA-I negative primary tumors: the road to
escape from T-cell responses. Curr Opin Immunol 51:123–132.
https​://doi.org/10.1016/j.coi.2018.03.006
11. McGranahan N, Rosenthal R, Hiley CT, Rowan AJ, Watkins
TBK, Wilson GA, Birkbak NJ, Veeriah S, Van Loo P, Herrero J,
Swanton C, Consortium TR (2017) Allele-specific HLA loss and
immune escape in lung cancer evolution. Cell 171(6):1259–1271
e1211. https​://doi.org/10.1016/j.cell.2017.10.001
Publisher’s Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
13