JID: CLLC

ARTICLE IN PRESS [mNS;April 11, 2021;3:24]

Original Study

PD-L1 Expression in Circulating Tumor Cells as a

Promising Prognostic Biomarker in Advanced
Non–small-cell Lung Cancer Treated with
Immune Checkpoint Inhibitors
Filippo G. Dall’Olio,1 Francesco Gelsomino,1 Nicole Conci,1 Laura Marcolin,2
Andrea De Giglio,1 Giada Grilli,1 Francesca Sperandi,1 Francesca Fontana,3
Mario Terracciano,3 Benedetta Fragomeno,1 Nastassja Tober,1 Giulia Manferrari,4
Stefano Brocchi,2 Rita Golfieri,2 Michelangelo Fiorentino,5 Andrea Ardizzoni1
Abstract
Biomarkers for immunotherapy represent a clinical need. Circulating tumor cells (CTCs) are associated with a
worse outcome. This is a prospective single-center cohort study enrolling 39 patients with non–small-cell lung
cancer. A median overall survival of 2.2, 3.7, and 16 months was observed in patients with negative CTC, positive
CTC, and no CTC detectable, respectively. No correlation was found between PD-L1 expression on CTCs and
on tumor tissue. CTCs were correlated with baseline tumor size. PD-L1 on CTCs is a promising biomarker.
Background: Circulating tumor cells (CTCs) are a promising source of biological information in cancer. Data correlating
PD-L1 expression in CTCs with patients’ response to immune checkpoint inhibitors (ICIs) in non–small-cell lung cancer
(NSCLC) are still lacking. Methods: This is a prospective single-center cohort study enrolling patients with advanced
NSCLC. CTCs were identified and counted with the CellSearch system. PD-L1 expression on CTCs was assessed
with phycoerythrin-conjugated anti-human PD-L1 antibody, clone MIH3 (BioLegend, USA). Primary endpoint was the
correlation between the CTCs PD-L1 expression and overall survival (OS). Among secondary objectives, we evaluated
the correlation between PD-L1 expression on CTCs and matched tumor tissue and the correlation of CTC number and
baseline tumor size (BTS). Results: Thirty-nine patients treated with anti-PD-1/PD-L1 agents as second- or third-line
therapy were enrolled. Patients were divided into 3 groups: no CTC detectable (CTCnull, n = 15), PD-L1 positive CTC
(CTCpos, n = 13), and PD-L1 negative CTC (CTCneg, n = 11). Median OS in patients with CTCneg was 2.2 months,
95% confidence interval (CI), 0.8-3.6 (reference) versus 3.7 months, 95% CI, 0.1-7.5 (hazard ratio [HR] 0.33; 95% CI,
0.13-0.83; P = .019) in patients with CTCpos versus 16.0 months, 95% CI, 2.2-29.8 (HR 0.17; 95% CI, 0.06-0.45; P<
.001) in patients with CTCnull. No correlation was found between PD-L1 expression on CTCs and on tumor tissue. CTC
number was correlated with BTS. Conclusion: PD-L1 expression on CTCs is a promising biomarker in patients with
NSCLC treated with ICIs. Further validation as predictive biomarker is needed.

Clinical Lung Cancer, Vol. 000, No.xxx, 1–9 © 2021 Elsevier Inc. All rights reserved.
Keywords: Advanced cancer, NSCLC, CTCs, PD-1, Predictive, Immune Checkpoint

1
Division of Oncology, IRCCS Azienda Ospedaliero-Universitaria di Bologna, Bologna, Italy
2
Department of Radiology, IRCCS Azienda Ospedaliero-Universitaria di Bologna, Bologna, Italy
3
Menarini Silicon Biosystems Spa, Bologna, Italy
4
Department of Genetics, Environment, and Evolution (GEE), University College London, London, United Kingdom
5
Department of Experimental, Diagnostic, and Specialty Medicine, University of Bologna, Bologna, Italy
Submitted: Jan 29, 2021; Revised: Feb 23, 2021; Accepted: Mar 4, 2021; Epub: xxx
Address for correspondence: Filippo Gustavo Dall’Olio, MD, IRCCS Azienda Ospedaliero-Universitaria di Bologna, Division of Oncology, Via Pietro Albertoni, 15, 40138 Bologna,
Italy.
E-mail contact: filippogdallolio@gmail.com

1525-7304/$ - see front matter © 2021 Elsevier Inc. All rights reserved.
https://doi.org/10.1016/j.cllc.2021.03.005 Clinical Lung Cancer 2021 1
Please cite this article as: Filippo G. Dall’Olio et al, PD-L1 Expression in Circulating Tumor Cells as a Promising Prognostic Biomarker in Advanced
Non–small-cell Lung Cancer Treated with Immune Checkpoint Inhibitors, Clinical Lung Cancer, https://doi.org/10.1016/j.cllc.2021.03.005
 JID: CLLC
ARTICLE IN PRESS
PD-L1 Expression in Circulating Tumor
Introduction
The unprecedented benefits from immune checkpoint inhibitors
(ICI) have revolutionized the treatment of patients affected by
non–small-cell lung cancer (NSCLC) without targetable driver
mutations,1-4 which represent the majority of NSCLC cases,
especially in western populations.5 Unfortunately, ICI proves to be
effective only in a limited number of this subset of patients, with
rates of long-term survivorship in unselected, pretreated patients
ranging from 22% to 37% at 2 years, 16% to 19% at 3 years, and
16% at 5 years.6
PD-L1 expression assessed by immunohistochemistry (IHC)
represents the only validated predictive marker of benefit from
ICI therapy. Nevertheless, unanticipated objective responses in
“negative” and disease progressions in “strongly positive” PD-L1
cases are also observed, questioning the predictive value of PD-L1.
In addition, the combination of ICI with chemotherapy has now
become the standard of care also in first-line therapy, regardless of
PD-L1 status.7 , 8
In fact, PD-L1 predictive role is severely impaired by the high
molecular heterogeneity both across different anatomic regions and
within single cancer-tissue samples itself.9 , 10 Finally, PD-L1 IHC
requires histological samples, and it is not validated on cytologic
smears, which are often the only available tissue samples for the
analysis. For this reason, the field of biomarkers research urges the
exploration of alternative PD-L1 indicators.
Circulating tumor cells (CTCs) are a promising source of valuable
information in cancer prognosis as they are both easily obtained,
presumably less impaired by tumor heterogeneity, and at the same
time can retain cancer-specificity and reflect the temporal evolution
of the tumor. Nevertheless, the isolation process of CTCs is not yet
standard practice in every clinical laboratory, as the technology is
not widespread and requires specific expertise.
Still, published data are encouraging, reporting an independent
prognostic significance of the number of detectable CTC,11-14 albeit
optimal criterion of positivity in CTC detection is still debated (at
least 1, 2, or 5 CTCs per 7.5 mL of blood), as well as the gold
standard detection assay, as different methods of CTC detection are
available. Broadly, they can be divided between an epithelial cell
adhesion molecule (EpCAM)-dependent (eg, CellSearch system)
and an EpCAM-independent approach (ligand-targeted polymerase
chain reaction, Cyttel, isolation by size of epithelial tumor cells
(ISET), and telomerase-based methods).15 , 16
Whichever the chased method for CTC enumeration, the
prognostic validity of CTC count is widely recognized for breast,17
colorectal,18 and prostate cancer,19 whereas data on the application
of CTCs as PD-L1 expression readout are still scarce in the context
of lung cancer immunotherapy, and studies looking at PD-L1 detec-
tion in CTCs report conflicting results.20
The intent of this study was to investigate the significance of CTC
presence and PD-L1 expression on CTCs as biomarker in pretreated
patients undergoing ICI therapy and the reliability of PD-L1 expres-
sion on CTC as compared with its tissue counterpart. We decided
to use EpCAM-dependent CellSearch system, as it is the only U.S.
Food and Drug Administration (FDA)-approved method for CTC
enumeration at present.
2 Clinical Lung Cancer 2021
Please cite this article as: Filippo G. Dall’Olio et al, PD-L1 Expression in Circulating Tumor Cells as a Promising Prognostic Biomarker in Advanced
Non–small-cell Lung Cancer Treated with Immune Checkpoint Inhibitors, Clinical Lung Cancer, https://doi.org/10.1016/j.cllc.2021.03.005
[mNS;April 11, 2021;3:24]
Materials and Methods
Patients
This was a single-center prospective cohort study. We included
consecutive patients with advanced NSCLC who were treated
in second- or third-line with single-agent ICI (nivolumab,
pembrolizumab, or atezolizumab) at Sant’Orsola Malpighi Univer-
sity Hospital between the September 20, 2018 and February 2,
2020. The end of follow-up period was October 1, 2020. All
patients signed an informed consent for the study.
CTC Detection and CTCPD-L1 Assessment
Blood samples were collected within 1 week from treatment initi-
ation in CellSave tubes (Menarini Silicon Biosystems, Bologna,
Italy) and processed within 96 hours. The analysis was outsourced
to an external laboratory.
CTCs were identified and counted with the CellSearch system
(Menarini Silicon Biosystems, Huntington Valley, PA), following
manufacturer directions. According to the FDA-approved defini-
tion, CTCs should be positive for the epithelial cell adhesion
molecule (EpCAM), cytokeratin (CK) identified by the clones C11
and A.53B/A2 and nuclear stain DAPI, and negative for leukocyte
marker CD45. CTCs counts were reported per 7.5 mL of blood.
CTCs PD-L1 assessment was performed with phycoerythrin-
conjugated anti-human PD-L1 antibody, clone MIH3 (BioLegend,
USA) on isolated CTCs.
Matched Primary Tumor Tissue
Tissue specimens were obtained at diagnosis and fixed in 4%
buffered formalin for 8 to 24 hours and embedded in paraffin.
Tissue sections of 3 μm were freshly cut from each formalin-fixed
paraffin embedded (FFPE) block before starting IHC.
IHC was performed with the automated Ventana BenchMark
ULTRA System (Ventana Medical System, Tucson, AZ), using the
OptiView DAB IHC Detection Kit (Ventana). PD-L1 reactivity was
assessed by IHC on FFPE samples, using the rabbit monoclonal
anti-PD-L1 (clone SP263; Ventana).
To demarcate the presence of tumor-infiltrating macrophages that
frequently overexpress PD-L1, a double staining with anti-PD-L1
(SP263) and anti-CD68 (KP-1, prediluted; Ventana) was adopted
on tissue samples.
In histological samples, PD-L1 expression was evaluated only in
cancer cells and at least 100 tumor cells per samples were scored.
Viable cancer cells were considered PD-L1 positive in the presence
of membrane staining of any intensity, whereas cytoplasmic staining
was considered positive only if there was concomitant membrane
staining.
Outcomes
The overarching endpoint of our work was to evaluate the corre-
lation between CTC abundance and PD-L1 expression on CTC
with overall survival (OS) in patients with NSCLC treated with
ICIs. In parallel, we sought to explore the correlation between PD-
L1 expression on CTCs and PD-L1 expression on matched tumor
tissue, the correlation of CTC count and PD-L1 expression on CTC
with progression-free survival (PFS), overall response rate (ORR),
JID: CLLC
ARTICLE IN PRESS
disease control rate (DCR), and durable clinical benefit (DCB),21
defined as alive and without disease progression (complete response,
partial response, or stable disease) at 6 months. Baseline tumor size
(BTS) was assessed by experienced radiologists and defined as the
sum of longest diameter (shortest for lymph nodes) of all target
lesion per RECIST 1.1 criteria (5 lesion maximum, up to 2 per
organ). Response was evaluated with computed tomography (CT)
scans performed each 8 to 12 weeks by experienced radiologists as
well using RECIST 1.1 criteria.
Statistical Analysis
REMARK guidelines (Reporting recommendations for tumor
MARKer prognostic studies) were followed throughout the
planning, analysis, and reporting of this study.22 Patients were
divided into 3 groups to set up an exploratory score: patients
without CTCs (CTCnull), patients with at least 1 PD-L1 positive
CTC (CTCpos), and patients with detectable CTCs but without
any PD-L1 positive CTC (CTCneg).
Patient clinical variables were compared using the χ 2 test or the
Fisher exact test for discrete variables. The concordance between
positivity of PD-L1 on CTCs (at least 1 PD-L1 cell isolated from
blood sample) and on tissue (PD-L1 ≥ 1%) was calculated with
Cohen’s k.
Clinical and pathological information was summarized using
summary statistics and correlated with unpaired t-test, and the
Wilcoxon sign-rank test when appropriate. OS and PFS were
estimated using the Kaplan-Meier method. Median follow-up was
calculated with reverse Kaplan-Meyer method.
Cox proportional hazard model was used to evaluate factors
independently associated with OS and PFS, whereas logistic regres-
sion was used for ORR and DCR. Variables included in the
final multivariate model were selected according to their clinical
relevance and statistical significance in a univariate analysis (P≤
.10). The multivariate model was designed using the backward
stepwise method. Internal validation of the final multivariate model
for OS, PFS, and for ORR was performed with a bootstrap sample
procedure (n = 1000 samples). The performance of the final model
was further quantified by the Harrell C index and validated with
bootstrap resampling procedure to calculate bias corrected C index.
The P value was considered significant when inferior to .05.
Statistical analysis was performed using the Statistical Package for
the Social Sciences (SPSS) program version 25.0 (IBM Corporation,
Armonk, NY).
Results
A total of 39 patients treated with anti-PD-1 or PD-L1 as second-
or third-line therapy were prospectively enrolled. The median age
was 68 years (range, 53-83); 15 were women; most of them had
nonsquamous histology (n = 30); an ECOG PS (Eastern Cooper-
ative Oncology Group Performance Status Scale) 0-1 (n = 30);
derived neutrophil to lymphocyte ratio (dNLR) < 3 (n = 22); and
had no liver (n = 27), bone (n = 27), and brain (n = 32) metastasis;
and 14 had lactate dehydrogenase (LDH) greater than upper limit of
normality (ULN) (Table 1). Median of the follow-up duration was
12.0 months (95% confidence interval [CI], 5.2-18.8). Detection of
PD-L1 expression on tissue could be assessed in 32 patients, ranging
Please cite this article as: Filippo G. Dall’Olio et al, PD-L1 Expression in Circulating Tumor Cells as a Promising Prognostic Biomarker in Advanced
Non–small-cell Lung Cancer Treated with Immune Checkpoint Inhibitors, Clinical Lung Cancer, https://doi.org/10.1016/j.cllc.2021.03.005
[mNS;April 11, 2021;3:24]
Filippo G. Dall’Olio et al
from 0% to 90%. CTCs were detected in 24 of the 39 patients
(62%), ranging from 1 to 59 cells per 7.5 mL of blood (Table 2).
Median BTS was 88.5 cm (standard deviation [SD] 55.8).
Patients were divided into 3 groups: no CTC detectable
(CTCnull, n = 15), ≥ 1 PD-L1 positive CTC (CTCpos, n = 13),
and CTC PD-L1 negative (CTCneg, n = 11). CTC presence
and PD-L1 expression were borderline correlated with gender
(P = .056). Moreover, BTS was correlated with the number of
CTC detected (r2 = 0.435, P = .008). Other characteristics were
balanced.
Concordance Between PD-L1 Positivity on Tissue and on
CTC
Our data showed no correlation between PD-L1 positivity on
tissue and on CTC (k = 0.292, P = .176). Sensitivity and specificity
were 69.2% and 60.0%, respectively, with a positive predictive value
of 69.2% and a negative predictive value of 60.0%. The mean PD-
L1 tumor proportion score (TPS) on tissue was 36.5% for CTCnull
(SD 38.5), 17.6% for CTCpos (SD 27.1), and 8.5% on CTCneg
(SD 12.9), without a significant correlation (P= .083). Similarly,
no significant difference was observed in terms of PD-L1 TPS on
tissue between those with CTCpos and CTCneg (P = .219). None
of CTCneg had PD-L1 TPS ≥ 50% (Figure 1).
Overall Survival
Median OS of the entire population was 4.2 months (95% CI,
2.3-6.1). At the time of the analysis, 29 patients (74% of total) had
died.
Median OS in patients with CTCneg was 2.2 months, 95% CI,
0.8-3.6 (reference), versus 3.7 months, 95% CI, 0.1-7.5 (hazard
ratio [HR], 0.33; 95% CI, 0.13-0.83; P = 0.019) in patients with
CTCpos versus 16.0 months, 95% CI, 2.2-29.8 (HR, 0.17; 95%
CI, 0.06-0.45; P< .001) in patients with CTCnull. There was no
difference between CTCpos and CTCnull (HR, 1.96; 95% CI,
0.77-4.98; P = .157) (Figure 2).
We defined an exploratory prognostic score based on the detec-
tion of CTCs and PD-L1 positivity (0 = no CTC, 1 = CTC PD-
L1pos, 2 = CTC PD-L1neg).
Multivariate analysis including PD-L1 positivity on CTCs
together with other well-established prognostic factors as ECOG PS,
liver and bone metastasis, LDH > ULN, and dNLR ≥ 3 confirmed
the independent prognostic value of PD-L1 expression on CTCs
(Table 3).
C index for the model comprising ECOG PS 2, dNLR ≥ 3, and
LDH > ULN was 0.757 (standard error [SE], 0.087), bias corrected
C index was 0.710, P< .001. When we added CTC score to the final
model, the C index increased to 0.867 (SE, 0.051), bias corrected C
index 0.821, P< .001.
Progression-Free Survival
PFS was calculated on a total of 38 patients as 1 was lost to follow-
up before first CT scan.
Median PFS was 2.6 months (95% CI, 2.0-3.3).
Median PFS in CTCneg patients was 1.9 months, 95% CI, 1.2-
2.6 (reference) versus 2.4 months, 95% CI, 0.6-4.2 (HR, 0.36;
95% CI, 0.15-0.87; P = .024) in CTCpos patients versus 3.1,
Clinical Lung Cancer 2021 3
 JID: CLLC
ARTICLE IN PRESS [mNS;April 11, 2021;3:24]

PD-L1 Expression in Circulating Tumor

Table 1 Clinical-Pathological Variables for all the Patients Enrolled in this Study

Clinical-Pathological Variables Total CTCnull CTCpos CTCneg P

ECOG PS 0-1 30 13 9 8 .511
(77%) (87%) (69%) (73%)
2 9 2 4 3
(23%) (13%) (31%) (27%)
dNLR <3 22 10 4 8 .332
(56%) (67%) (31%) (73%)
≥3 12 4 5 3
(31%) (27%) (38%) (27%)
NA 5 1 4 0
(13%) (6%) (31) (0%)
LDH > ULN 14 6 4 4 .821
≤ ULN 15 8 3 4
NA 10 1 6 3
Brain No 32 13 10 9 .657
metastasis (82%) (87%) (77%) (82%)
Yes 6 2 3 1
(15%) (13%) (23%) (9%)
NA 1 0 0 1
(3%) (0%) (0%) (9%)
Bone No 27 10 10 7 .921
metastasis (69%) (67%) (77%) (64%)
Yes 10 4 3 3
(26%) (27%) (23%) (27%)
NA 2 1 0 1
(5%) (6%) (0%) (9%)
Liver No 27 10 9 8 .947
metastasis (69%) (67%) (69%) (73%)
Yes 12 5 4 3
(31%) (33%) (31%) (27%)
Sex Male 24 7 7 10 .056
(62%) (47%) (54%) (91%)
Female 15 8 6 1
(38%) (53%) (46%) (9%)
Histotypes Nonsquamous 30 13 11 6 .114
(77%) (87%) (85%) (55%)
Squamous 9 2 2 5
(23%) (13%) (15%) (45%)
Age Mean (SD) 68 (11) 72 (15) 68 (7) 70 (9) .659
PD-L1 Mean (SD) 16.0 (25.9) 36.5 (38.5) 17.6 (27.1) 8.5 (12.9) .083
expression
Total Nr pts 39 15 13 11

The variables are reported for patients with no CTC detectable (CTCnull), CTC detectable with at least 1 CTC PD-L1 positive (CTCpos), and CTC detectable with none of them PD-L1 positive (CTCneg),
respectively.
Abbreviations: CTC = circulating tumor cells; dNLR = derived neutrophil to lymphocyte ratio; ECOG PS = Eastern Cooperative Oncology Group Performance Status Scale; LDH = lactate dehydrogenase;
NA = not available; ULN = upper limit of normality; Nr pts = number of patients; SD = standard deviation.

95% CI, 1.8-4.3 (HR, 0.37; 95% CI, 0.15-0.95; P = .040) in
CTCnull patients (Figure 3). There was no difference between
CTCpos and CTCnull (HR, 1.03; 95% CI, 0.43-2.45; P = .950).
Multivariate analysis including PD-L1 positivity on CTCs together
with other previously reported prognostic factors as ECOG PS,
presence of liver and bone metastasis, and dNLR ≥ 3, confirmed
the independent prognostic value of PD-L1 expression on CTCs
(Table 4).
Moreover, 5 out of 15 in the CTCnull group experienced DCB
versus 3 out of 14 in CTCpos group and none of the 11 patients in
the CTCneg group.
Radiologic Response
Of the 26 patients evaluable for radiologic response, 5 out or 13
(38%) in CTCnull group achieved an objective response versus 3

4 Clinical Lung Cancer 2021

Please cite this article as: Filippo G. Dall’Olio et al, PD-L1 Expression in Circulating Tumor Cells as a Promising Prognostic Biomarker in Advanced
Non–small-cell Lung Cancer Treated with Immune Checkpoint Inhibitors, Clinical Lung Cancer, https://doi.org/10.1016/j.cllc.2021.03.005
JID: CLLC
ARTICLE IN PRESS [mNS;April 11, 2021;3:24]

Filippo G. Dall’Olio et al
Table 2 Overall Survival and Best Response of Patients According to the Number of CTCs and Number of CTCs PD-L1 Positive

Patients N OS CTC Count Number of CTC PD-L1 Positive Best Response

1 4.2 3 0 PD
2 0.8 2 0 NA
3 2.2 2 0 PD
4 3.7 5 5 PD
5 0.9 2 1 NA
6 1.7 1 0 PD
7 22.4 0 0 SD
8 5.0 0 0 PR
9 22.1 1 1 PR
10 3.1 0 0 PD
11 0.7 4 3 NA
12 0.3 1 0 NA
13 3.7 1 0 PD
14 16.0 0 0 PD
15 5.4 1 1 NA
16 4.3 1 0 PD
17 11.6 0 0 PR
18 18.0 2 2 PR
19 3.4 3 1 SD
20 0.1 0 0 NA
21 1.0 18 0 NA
22 17.1 3 3 PR
23 4.0 0 0 PD
24 9.9 23 14 PD
25 16.8 0 0 SD
26 9.5 0 0 PR
27 6.3 0 0 PR
28 0.4 59 3 NA
29 10.7 2 2 PD
30 12.0 0 0 NA
31 4.5 37 18 NA
32 10.6 0 0 PD
33 2.9 0 0 SD
34 1.1 0 0 NA
35 3.2 0 0 PD
36 1.5 6 1 NA
37 2.8 4 0 NA
38 3.2 1 0 NA
39 1.5 2 0 NA

CTC = circulating tumor cells; NA = not assessed; OS = overall survival; PD = progressive disease; PR = partial response; SD = stable disease.

out of 7 (43%) in CTCpos group and 0 out of 6 (0%) in CTCneg Discussion

group P = 0.178.
Similarly, when considering DCR, 8/13 (62%) patients achieved
at least a disease stabilization in CTCnull group versus 4/7
(57%) in CTCpos group and none (0%) in CTCneg group,
P = 0.035.
In this study we analyzed the prognostic impact of CTCs
presence, as assessed by the CellSearch system, and their PD-L1
expression in platinum pretreated patients with NSCLC eligible for
single-agent ICI therapy. We also analyzed the correlation of PD-L1
expression between CTCs and tissue biopsy.

Clinical Lung Cancer 2021 5

Please cite this article as: Filippo G. Dall’Olio et al, PD-L1 Expression in Circulating Tumor Cells as a Promising Prognostic Biomarker in Advanced
Non–small-cell Lung Cancer Treated with Immune Checkpoint Inhibitors, Clinical Lung Cancer, https://doi.org/10.1016/j.cllc.2021.03.005
 JID: CLLC
ARTICLE IN PRESS [mNS;April 11, 2021;3:24]

PD-L1 Expression in Circulating Tumor

Figure 1 Scatter plot of PD-L1 expression in matched tissue sample for no CTC detectable (CTCnull), positive CTC (CTCpos), and
negative CTC (CTCneg). CTC = circulating tumor cells.

Figure 2 Kaplan-Meyer curves for overall survival analysis of patients with no CTC detectable (CTCnull), CTC detectable with at
least 1 CTC PD-L1 positive (CTCpos), and CTC detectable with none of them PD-L1 positive (CTCneg).
CTC = circulating tumor cells.

We found that CTCs presence and PD-L1 expression correlated
with outcomes in terms of OS and PFS, as well as DCR. Impor-
tantly, we proposed a CTC score useful to predict the prognosis
of patients treated with immunotherapy at baseline, which proved
to identify a subset of patients unlikely to benefit from ICI alone.
This score appeared to retain an independent prognostic value in
multivariate analysis comprising also other well-established prognos-
tic variables such as dNLR, bone metastasis, LDH, and ECOG
PS > 1.23-27
The CTCs detection rate in our study (63%) was slightly higher
than the previously reported values, which ranged approximately
between 20% and 50% and is usually lower in studies using an
EpCAM-dependent approach.11 This is a double-edged sword, as
the absence of CTCs detection retains a prognostic value by itself,
but the unsatisfyingly low detection rate does not allow to perform
further analysis such as PD-L1 CTC assessment.
As already described, CTC number was correlated with progno-
sis (both generally28 and specifically in patients treated with ICI)29
and BTS, with larger tumors shedding an higher amount of CTC.
However, as shown by the multivariate analysis, our score could
add substantial information and is not simply a surrogate marker
of tumor burden.
In our article, PD-L1 expression on CTC provided important
prognostic information too. The predictive role of PD-L1 expres-
sion on CTCs in patients receiving ICI is uncertain, as reports are
discordant and affected by a considerable degree of heterogeneity in
conduction. Also, they generally showed a positive prognostic value
for PD-L1 expression on CTCs.30-32

6 Clinical Lung Cancer 2021

Please cite this article as: Filippo G. Dall’Olio et al, PD-L1 Expression in Circulating Tumor Cells as a Promising Prognostic Biomarker in Advanced
Non–small-cell Lung Cancer Treated with Immune Checkpoint Inhibitors, Clinical Lung Cancer, https://doi.org/10.1016/j.cllc.2021.03.005
JID: CLLC
ARTICLE IN PRESS [mNS;April 11, 2021;3:24]

Filippo G. Dall’Olio et al
Table 3 Univariate and Multivariate Analysis for Overall Survival

Univariate Multivariate
Variable HR 95% CI P HR 95% CI P
CTC score 2.45 1.39 4.01 .001 2.08 1.08 4.03 .03
ECOG PS (2 vs. 0-1) 3.21 1.31 7.87 .011
Liver metastasis (presence vs. absence) 1.45 0.62 3.36 .392
Bone metastasis (presence vs. absence) 2.74 1.13 6.66 .026 2.189 .669 7.160 .195
dNLR (≥ 3 vs. < 3) 5.57 2.15 14.41 <.001 6.45 1.98 21.01 .002
PD-L1 expression on tissue 0.99 0.97 1.01 .159
Brain metastasis (presence vs. absence) 0.72 0.21 2.41 .591
Histology (squamous vs. nonsquamous) 2.45 1.08 5.58 .032 2.75 0.93 8.12 .067
LDH > ULN 2.53 0.998 6.405 .051 1.98 0.69 5.68 .203

Abbreviations: CI = confidence interval; CTC = circulating tumor cells; dNLR = derived neutrophil to lymphocyte ratio; ECOG PS = Eastern Cooperative Oncology Group Performance Status Scale;
HR = hazard ratio; LDH = lactate dehydrogenase; ULN = upper limit of normality.

Figure 3 Kaplan-Meyer curves for progression-free survival analysis of patients with no CTC detectable (CTCnull), CTC
detectable with at least 1 CTC PD-L1 positive (CTCpos), and CTC detectable with none of them PD-L1 positive
(CTCneg). CTC = circulating tumor cells.

Conversely, other studies investigated the prognostic validity of
PD-L1 expression on CTCs in patients receiving chemotherapy,
finding instead that PD-L1+ CTCs were borderline33 or signifi-
cantly associated with worse outcome.34
Interestingly, our article shows no correlation between PD-L1
expression assessed on CTCs and on tissue biopsy. This piece of
evidence is consistent with other reports. Guibert et al.30 found no
correlation between tissue and CTC PD-L1 expression (r = 0.04,
P = .77). Similarly, the article by Koh et al.35 reported no corre-
lation (R2 0.0034). In both these articles, the researchers used an
EpCAM-independent system for CTCs detection. Janning et al.36
used an EpCAM-dependent system (CellSearch) and likewise found
no correlation. Our article therefore confirms the scarce correlation
between PD-L1 status on CTCs and on matched tumor tissue and
raises doubts of the predictive value of PD-L1 based on a single
tissue biopsy. The reason for the absence of correlation could also
be explained in the low detection rate. An interesting finding is that
none of the CTCneg patients had a TPS ≥ 50% on the matched
tumor tissue. This point deserves further investigations in a wider
population.
Among the limitations of our work, the high rate of undetected
CTCs increases the possible false-negative rate and underlines the
need of improvement in the detection technique.
Moreover, an EpCAM-based approach fails to capture EpCAM-
negative cells that could result from epithelial-to-mesenchymal
transition.37 Finally, considering the low number of patients
enrolled, the question if the proportion of PD-L1 positive cells can
have an impact on the outcome still needs to be answered.
To the best of our knowledge, however, this is the first prospective
study to purpose a prognostic score based both on CTC detection
with the CellSearch system and PD-L1 expression. This score retains
independent prognostic validity. Moreover, considering the negative

Clinical Lung Cancer 2021 7

Please cite this article as: Filippo G. Dall’Olio et al, PD-L1 Expression in Circulating Tumor Cells as a Promising Prognostic Biomarker in Advanced
Non–small-cell Lung Cancer Treated with Immune Checkpoint Inhibitors, Clinical Lung Cancer, https://doi.org/10.1016/j.cllc.2021.03.005
 JID: CLLC
ARTICLE IN PRESS [mNS;April 11, 2021;3:24]

PD-L1 Expression in Circulating Tumor

Table 4 Univariate and Multivariate Analysis for Progression-Free Survival

Univariate Multivariate
Variable HR 95% CI P HR 95% CI P
CTC score 1.62 1.02 2.58 .043 1.64 1.00 2.70 .05
ECOG PS (2 vs. 0-1) 2.53 1.12 5.73 .025 1.75 0.67 4.57 .252
Liver metastasis (presence vs. absence) 1.67 0.79 3.50 .178
Bone metastasis (presence vs. absence) 2.59 1.14 5.91 .023 1.75 0.61 5.00 .298
dNLR (≥ 3 vs. < 3) 7.30 2.82 18.90 <.001 8.01 2.83 22.68 <.001
PD-L1 expression on tissue 0.99 0.97 1.01 .186
Brain metastasis (presence vs. absence) 1.72 0.65 4.56 .273
LDH > ULN 2.21 .68 7.18 .198

Abbreviations: CI = confidence interval; CTC = circulating tumor cells; dNLR = derived neutrophil to lymphocyte ratio; ECOG PS = Eastern Cooperative Oncology Group Performance Status Scale;
HR = hazard ratio; LDH = lactate dehydrogenase; ULN = upper limit of normality.

impact of PD-L1 expression on CTC in patients treated with differ-
ent class of agents, it can be speculated that in turn PD-L1 CTCs
detection could imply predictive value for immunotherapy benefit,
although this hypothesis requires further validation in a different set
of data with a control group to be confirmed.
Notably, the independent value of PD-L1 expression on CTC
with respect to PD-L1 expression on tissue suggests that this analy-
sis could play a pivotal role in selecting patients to treat with ICI.
Moreover, future studies could explore a score that includes CTC
PD-L1 assessment, as well as other prognostic factors such as ECOG
PS and dNLR to develop a model that could improve prognostic
accuracy and also be useful to deal with the complexity of a single
patient’s case.
Conclusions
Our article confirms the prognostic validity of CTCs detection in
blood sample and suggest that PD-L1 assessment on CTC could add
a substantial prognostic value for patients with NSCLC receiving
ICI therapy. We found no correlation between PD-L1 expression
on tissue and on CTCs. Further analysis on a larger set of data is
warranted to investigate the predictive value of our score.
Clinical Practice Points
• Predictive biomarkers for immunotherapy in NSCLC still repre-
sent an unmet clinical need.
• CTCs have been associated with a worse outcome.
• In our study, CTC abundance confirmed to be a negative
prognostic factor.
• PD-L1 expression on CTCs correlates with a better outcome.
• No correlation was found between PD-L1 expression on CTCs
and on tumor tissue.
Disclosure
This research received its core funding from the nonprofit organi-
zation “AIRC - Fondazione AIRC per la Ricerca sul Cancro” (inves-
tigator grant 19026).
Andrea Ardizzoni received grants from BMS and Celgene;
personal fees from BMS, MSD, Eli Lilly, Boehringer, Pfizer, and
Celgene. Francesca Fontana and Mario Terraciano are employees
of Silicon Biosystem. The other authors do not report any relevant
conflicts of interest.
References
1. Borghaei H, Paz-Ares L, Horn L, et al. Nivolumab versus docetaxel in advanced
nonsquamous non–small-cell lung cancer. N Engl J Med. 2015;373:1627–1639.
2. Brahmer J, Reckamp KL, Baas P, et al. Nivolumab versus docetaxel in advanced
squamous-cell non–small-cell lung cancer. N Engl J Med. 2015;373:123–135.
3. Fehrenbacher L, Spira A, Ballinger M, et al. Atezolizumab versus docetaxel for
patients with previously treated non-small-cell lung cancer (POPLAR): a multicen-
tre, open-label, phase 2 randomised controlled trial. Lancet. 2016;387:1837–1846.
4. Reck M, Rodríguez-Abreu D, Robinson AG, et al. Pembrolizumab versus
chemotherapy for PD-L1–positive non–small-cell lung cancer. N Engl J Med.
2016;375:1823–1833.
5. Dall’Olio FG, Conci N, Rossi G, et al. Comparison of sequential testing and next
generation sequencing in advanced lung adenocarcinoma patients–a single centre
experience. Lung Cancer. 2020;149:5–9.
6. Nadal E, Massuti B, Dómine M, García-Campelo R, Cobo M, Felip E.
Immunotherapy with checkpoint inhibitors in non-small cell lung cancer: insights
from long-term survivors. Cancer Immunol Immunother. 2019;68:341–352.
7. Gandhi L, Rodríguez-Abreu D, Gadgeel S, et al. Pembrolizumab plus chemother-
apy in metastatic non–small-cell lung cancer. N Engl J Med. 2018;378:2078–2092.
8. Paz-Ares L, Luft A, Vicente D, et al. Pembrolizumab plus chemotherapy for
squamous non–small-cell lung cancer. N Engl J Med. 2018;379:2040–2051.
9. McLaughlin J, Han G, Schalper KA, et al. Quantitative assessment of the
heterogeneity of PD-L1 expression in non-small-cell lung cancer. JAMA Oncol.
2016;2:46–54.
10. Munari E, Zamboni G, Marconi M, et al. PD-L1 expression heterogeneity in
non-small cell lung cancer: evaluation of small biopsies reliability. Oncotarget.
2017;8:90123–90131.
11. Krebs MG, Hou JM, Sloane R, et al. Analysis of circulating tumor cells in patients
with non-small cell lung cancer using epithelial marker-dependent and -indepen-
dent approaches. J Thorac Oncol. 2012;7:306–315.
12. Bayarri-Lara C, Ortega FG, Cueto Ladrón de Guevara A, et al. Circulating tumor
cells identify early recurrence in patients with non-small cell lung cancer undergo-
ing radical resection.. PLoS One. 2016;11.
13. Zhang Z, Xiao Y, Zhao J, et al. Relationship between circulating tumour cell count
and prognosis following chemotherapy in patients with advanced non-small-cell
lung cancer. Respirology. 2016;21:519–525.
14. Crosbie PAJ, Shah R, Krysiak P, et al. Circulating tumor cells detected in the
tumor-draining pulmonary vein are associated with disease recurrence after surgical
resection of NSCLC. J Thorac Oncol. 2016;11:1793–1797.
15. De Wit S, Van Dalum G, Lenferink ATM, et al. The detection of EpCAM+ and
EpCAM- circulating tumor cells. Sci Rep. 2015;5:12270.
16. Gabriel MT, Calleja LR, Chalopin A, Ory B, Heymann D. Circulating tumor cells:
a review of Non-EpCAM-based approaches for cell enrichment and isolation. Clin
Chem. 2016;62:571–581.
17. Cristofanilli M, Pierga JY, Reuben J, et al. The clinical use of circulating tumor cells
(CTCs) enumeration for staging of metastatic breast cancer (MBC): international
expert consensus paper. Crit Rev Oncol Hematol. 2019;134:39–45.
18. Tol J, Koopman M, Miller MC, et al. Circulating tumour cells early predict
progression-free and overall survival in advanced colorectal cancer patients treated
with chemotherapy and targeted agents. Ann Oncol Off J Eur Soc Med Oncol.
2010;21:1006–1012.
19. Di Nunno V, Gatto L, Santoni M, et al. Recent advances in liquid biopsy in
patients with castration resistant prostate cancer. Front Oncol. 2018;8:397.

8 Clinical Lung Cancer 2021

Please cite this article as: Filippo G. Dall’Olio et al, PD-L1 Expression in Circulating Tumor Cells as a Promising Prognostic Biomarker in Advanced
Non–small-cell Lung Cancer Treated with Immune Checkpoint Inhibitors, Clinical Lung Cancer, https://doi.org/10.1016/j.cllc.2021.03.005
JID: CLLC
ARTICLE IN PRESS
20. Tong B, Wang M. Circulating tumor cells in patients with lung cancer: develop-
ments and applications for precision medicine. Futur Oncol. 2019;15:2531–2542.
21. Rizvi NA, Hellmann MD, Snyder A, et al. Mutational landscape determines sensi-
tivity to PD-1 blockade in non-small cell lung cancer.. Science. 2015;348:124–128.
22. McShane LM, Altman DG, Sauerbrei W, et al. REporting recommendations for
tumour MARKer prognostic studies (REMARK). Br J Cancer. 2005;93:387–391.
23. Laura M, Edouard A, Roberto F, et al. Association of the lung immune prognos-
tic index with immune checkpoint inhibitor outcomes in patients with advanced
non-small cell lung cancer. JAMA Oncol. 2018;4:351–357.
24. Tumeh PC, Hellmann MD, Hamid O, et al. Liver metastasis and treatment
outcome with anti-PD-1 monoclonal antibody in patients with melanoma and
NSCLC. Cancer Immunol Res. 2017;5:417–424.
25. Dall’Olio FG, Abbati F, Facchinetti F, et al. CEA and CYFRA 21-1 as prognostic
biomarker and as a tool for treatment monitoring in advanced NSCLC treated with
immune checkpoint inhibitors. Ther Adv Med Oncol. 2020;12.
26. Dall’Olio FG, Maggio I, Massucci M, Mollica V, Fragomeno B, Ardizzoni A.
ECOG performance status ≥2 as a prognostic factor in patients with advanced
non small cell lung cancer treated with immune checkpoint inhibitors—a system-
atic review and meta-analysis of real world data. Lung Cancer. 2020;145:95–104.
27. Facchinetti F, Mazzaschi G, Barbieri F, et al. First-line pembrolizumab in advanced
non–small cell lung cancer patients with poor performance status. Eur J Cancer.
2020;130:155–167.
28. Lindsay CR, Blackhall FH, Carmel A, et al. EPAC-lung: pooled analysis of
circulating tumour cells in advanced non-small cell lung cancer. Eur J Cancer.
2019;117:60–68.
29. Tamminga M, de Wit S, Hiltermann TJN, et al. Circulating tumor cells in
advanced non-small cell lung cancer patients are associated with worse tumor
response to checkpoint inhibitors. J Immunother Cancer. 2019;7:173.
Please cite this article as: Filippo G. Dall’Olio et al, PD-L1 Expression in Circulating Tumor Cells as a Promising Prognostic Biomarker in Advanced
Non–small-cell Lung Cancer Treated with Immune Checkpoint Inhibitors, Clinical Lung Cancer, https://doi.org/10.1016/j.cllc.2021.03.005
[mNS;April 11, 2021;3:24]
Filippo G. Dall’Olio et al
30. Guibert N, Delaunay M, Lusque A, et al. PD-L1 expression in circulating tumor
cells of advanced non-small cell lung cancer patients treated with nivolumab. Lung
Cancer. 2018;120:108–112.
31. Khattak MA, Reid A, Freeman J, et al. PD-L1 expression on circulating tumor cells
may be predictive of response to pembrolizumab in advanced melanoma: results
from a pilot study. Oncologist. 2020;25:e520–e527.
32. Yue C, Jiang Y, Li P, et al. Dynamic change of PD-L1 expression on circulating
tumor cells in advanced solid tumor patients undergoing PD-1 blockade therapy..
Oncoimmunology. 2018;7.
33. Ilié M, Szafer-Glusman E, Hofman V, et al. Detection of PD-L1 in circulating
tumor cells and white blood cells from patients with advanced non-small-cell lung
cancer. Ann Oncol. 2018;29:193–199.
34. Kulasinghe A, Kapeleris J, Kimberley R, et al. The prognostic significance of circu-
lating tumor cells in head and neck and non-small-cell lung cancer. Cancer Med.
2018;7:5910–5919.
35. Koh Y, Yagi S, Akamatsu H, et al. Heterogeneous expression of programmed death
receptor-ligand 1 on circulating tumor cells in patients with lung cancer. Clin Lung
Cancer. 2019;20:270–277 e1.
36. Janning M, Kobus F, Babayan A, et al. Determination of PD-L1 expression
in circulating tumor cells of NSCLC patients and correlation with response to
PD-1/PD-L1 inhibitors. Cancers (Basel). 2019;11:835.
37. Grover PK, Cummins AG, Price TJ, Roberts-Thomson IC, Hardingham JE.
Circulating tumour cells: the evolving concept and the inadequacy of their enrich-
ment by EpCAM-based methodology for basic and clinical cancer research. Ann
Oncol. 2014;25:1506–1516.
Clinical Lung Cancer 2021 9