Fermat's Library | Stratification of PD1 blockade response in melanoma using pre- & post-treatment immunophenotyping of peripheral blood annotated/explained version.

Did you look at whether there was any correlation between the CD4:C...

Can you speculate on why you saw a difference in the frequency of P...

Do you think that responders are likely to have more antigen experi...

Do you think that these phenotypes would also been seen and predict...

Immunotherapy Advances, 2023, 3, 1–13

https://doi.org/10.1093/immadv/ltad001

Advance access publication 6 January 2023

Research Article

Stratiﬁcation of PD-1 blockade response in melanoma

using pre- and post-treatment immunophenotyping of

peripheral blood

Natalie M.Edner

1,*,

, ElisavetNtavli

, LinaPetersone

, Chun JingWang

, AstridFabri

AlexandrosKogimtzis

, VitalijsOvcinnikovs

, Ellen M.Ross

, FrankHeuts

, YassinElfaki

Luke P.Houghton

, TobyTalbot

, AmnaSheri

, AlexandraPender

, DavidChao

3,†

and

LucyS.K.Walker

1,†,

Institute of Immunity & Transplantation, University College London Division of Infection & Immunity, London, UK

Royal Cornwall Hospitals NHS Trust, Truro, UK

Department of Oncology, Royal Free London NHS Foundation Trust, London, UK

†

These authors contributed equally to this work.

Correspondence: Natalie M. Edner, Institute of Immunity & Transplantation, University College London Division of Infection & Immunity, Royal Free Campus,

London, NW3 2PP, UK. Email: natalie.edner.15@ucl.ac.uk

Summary

Efﬁcacy of checkpoint inhibitor therapies in cancer varies greatly, with some patients showing complete responses while others do not re-

spond and experience progressive disease. We aimed to identify correlates of response and progression following PD-1-directed therapy by

immunophenotyping peripheral blood samples from 20 patients with advanced malignant melanoma before and after treatment with the PD-1

blocking antibody pembrolizumab. Our data reveal that individuals responding to PD-1 blockade were characterised by increased CD8 T cell prolif-

eration following treatment, while progression was associated with an increase in CTLA-4-expressing Treg. Remarkably, unsupervised clustering

analysis of pre-treatment T cell subsets revealed differences in individuals that went on to respond to PD-1 blockade compared to individuals that

did not. These differences mapped to expression of the proliferation marker Ki67 and the costimulatory receptor CD28 as well as the inhibitory

molecules 2B4 and KLRG1. While these results require validation in larger patient cohorts, they suggest that ﬂow cytometric analysis of a relatively

small number of T cell markers in peripheral blood could potentially allow stratiﬁcation of PD-1 blockade treatment response prior to therapy initiation.

Graphical Abstract

Keywords: checkpoint inhibition, PD-1 blockade, immunotherapy, biomarker research

Received: August 17, 2022; Accepted: January 4, 2023

This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/),

which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.

Downloaded from https://academic.oup.com/immunotherapyadv/article/3/1/ltad001/6972738 by Queen Mary University of London user on 01 March 2023

2 N.M. Edner et al.

Abbreviations: CR: Complete responder; FCS: Foetal calf serum; MFI: Median uorescence intensity; PBMC: Peripheral blood mononuclear cell; PBS: Phosphate-

buffered saline; PCA: Principal component analysis; PD: Progressive disease; PR: Partial responder; SD: Stable disease; Tfr: Follicular regulatory T cells; TIL:

Tumour-inltrating lymphocyte; Treg: Regulatory T cells.

Introduction

Harnessing the power of the immune system in the form of

immunomodulatory drugs has revolutionised cancer therapy,

with checkpoint inhibitors targeting the coinhibitory PD-1

and CTLA-4 pathways now being the standard of care in the

treatment of many types of malignancies [1]. These therapies

aim to reinvigorate tumour-specic T cell responses and re-

lease tumour-inltrating lymphocytes (TILs) from tumour

suppression. In particular, antibodies blocking PD-1 or its li-

gand PD-L1 have caused a paradigm shift in cancer care with

an unprecedented 5-year overall survival of 34% for patients

with advanced melanoma treated with the PD-1 inhibitor

pembrolizumab [2].

Nevertheless, the clinical response following PD-1 pathway

interventions shows great heterogeneity and varies across in-

dividual patients and different types of tumours [3]. Therefore,

considerable effort has been directed towards identication of

biomarkers that stratify patient response following therapy.

For example, expression of PD-L1 on cells residing within the

tumour has been shown to have some positive correlation with

response, although patients with PD-L1-negative tumours can

still respond to PD-1 blockade [4, 5]. Furthermore, a higher

tumour mutational burden and abundance of neoantigens

expressed by tumour cells is favourable since it can facilitate

anti-tumour immune responses [6, 7]. Similarly, the presence

of TILs, particularly CD8 T cells, is associated with a better

response following PD-1 blockade [8, 9].

However, to predict patient response using these markers,

tumour biopsies, ideally taken at multiple sites, are required,

which is not always feasible, and robust biomarkers there-

fore still present an unmet clinical need in cancer immu-

notherapy. As an alternative, peripheral blood can provide

a snapshot of the systemic immune response and is easily

accessible. Tumour-specic T cells have been detected in the

circulation and T cell clones reinvigorated by PD-1 blockade

can be found in the blood as well as tumour inltrates

[10–12]. Relatively few studies have investigated potential

biomarkers for response to PD-1 blockade in ex vivo periph-

eral blood lymphocytes and those that have been completed

primarily focus on CD8 T cells [13–18]. Consequently, we

sought to identify immune correlates of clinical response fol-

lowing PD-1 blockade by immunophenotyping both CD4

and CD8 T cells in fresh peripheral blood samples. We

have previously performed immunophenotyping of circu-

lating lymphocytes in autoimmune settings where we have

identied biomarkers of response to immunotherapy [19,

20]. Drawing on this experience, a range of ow cytometry

panels were designed and applied to blood samples from

patients with advanced malignant melanoma before and

after anti-PD-1 therapy.

Methods and materials

Patients

Patients with advanced malignant melanoma of cutaneous

or mucosal origin were recruited at the Royal Free Hospital

NHS Foundation Trust as part of the PASIP research study

(NCT02909348). The protocol and consent document of

this trial were approved by the London - Bromley Research

Ethics Committee (16/LO/1296). All participants were above

the age of 18, provided written, informed consent and were

deidentied prior to analysis. Patient characteristics of the

study cohort are displayed in Supplementary Table S1. Patients

were treated with 2 mg/kg of pembrolizumab (humanised

monoclonal anti-PD-1, Keytruda, MSD) intravenously on

day 1 of each 21-day cycle until progression or unacceptable

toxicity developed. Blood samples were typically taken before

treatment and then 6 and 12 weeks after treatment initiation.

Assessment of patient response was performed prior to cycle

5 infusion by the treating physician according to RECIST ver-

sion 1.1 [21].

Sample preparation

Peripheral blood mononuclear cells (PBMCs) were isolated

from whole blood by density gradient centrifugation. In

brief, whole blood was diluted in phosphate-buffered saline

(PBS, Merck) and layered over Histopaque 1077 (Merck) in

LeucoSep 50 ml tubes (Greiner Bio-One). After centrifuga-

tion at 800 × g for 15 minutes with no brake, supernatant

above barrier was poured into new 50 ml Falcon tube and

centrifuged at 350 × g for 10 minutes to remove remaining

histopaque. Supernatant was discarded and pellet resuspended

in PBS + 2% foetal calf serum (FCS, Life Science Production).

Platelets were removed by centrifugation at 250 × g for 10

minutes. The resulting pellet was resuspended in PBS + 2%

FCS and 1 × 10

cells per ow cytometry panel were used for

subsequent ow cytometry staining.

Flow cytometry

PBMCs were surface stained with four panels of antibody

cocktails (see Supplementary Table S2) for 15 minutes at

37°C. For panel 1, cells were then incubated with streptavidin

APC for 10 minutes at 4°C. Cells were incubated with xable

viability dye eFluor 780 (Thermo Fisher Scientic) in PBS for

10 minutes at 4°C. For intracellular staining in panel 2, cells

were xed and permeabilized using the Foxp3/Transcription

Factor Staining Buffer Set (Thermo Fisher Scientic) and

incubated with an antibody cocktail containing anti-human

Ki67 Alexa Fluor 488 (clone: Ki-67, Biolegend), CTLA-4 PE

(clone: BNI3, BD Biosciences) and Foxp3 APC (clone: 236A/

E7, Thermo Fisher Scientic) for 30 minutes at 4°C. Samples

were acquired on a BD LSRFortessa (BD Biosciences) using

the BD FACSDiva software v.8.0.2 (BD Biosciences).

Data analysis

For manual analysis, ow cytometry data were analysed

using FlowJo v.10 (FlowJo LLC). For unsupervised clustering,

pregated live, singlet lymphocytes were preprocessed in R

v.4.0.2 as previously described [20] and the FlowSOM al-

gorithm as implemented in the Bioconductor package

CATALYST v.1.14.1 was used to identify CD4 and CD8 T

cell populations. FlowSOM clustering was then applied again

on these T cell populations and optimal number of clusters

was identied using delta area plots.

For tSNE projections, ow cytometry data were

downsampled to 3000 cells per sample and CATALYST was

Downloaded from https://academic.oup.com/immunotherapyadv/article/3/1/ltad001/6972738 by Queen Mary University of London user on 01 March 2023

Immune correlates of αPD-1 response in melanoma 3

used to compute tSNE embedding. Principal component anal-

ysis (PCA) was performed on scaled and centred data. Prior to

PCA, cluster correlation was calculated using Pearson’s r and,

for cluster pairs that were correlated with an r ≥ 0.95, one

cluster was randomly removed from analysis. Statistical anal-

ysis was performed using R. A two-tailed Student’s t test was

used to compare two unpaired means. Plots were produced

using the CRAN packages ggplot2 v.3.3.5, ggpubr v.0.4.0,

ggtext v.0.1.1, cowplot v.1.1.1, scico v.1.3.0, and circlize

v.0.4.13 and the Bioconductor packages ComplexHeatmap

v.2.6.2. Data cleaning and formatting was carried out using

CRAN packages tidyr v.1.1.4, reshape2 v.1.4.4, Rmisc v.1.5,

rstatix v.0.7.0, and lubridate v.1.8.0.

Data sharing statement

Deidentied individual participant data supporting the

ndings of this study are available from the corresponding

author upon request.

Results

Incomplete PD-1 blockade on CD4 T cells in

non-responders

In order to investigate the effect of PD-1 blockade on pe-

ripheral immune cells, blood samples from 20 patients

with advanced malignant melanoma were collected in the

PASIP study. In this study, patients received treatment with

pembrolizumab every 3 weeks and blood samples were

processed at baseline and 6 and 12 weeks following treatment

initiation (Fig. 1A). Response was assessed at the pre-cycle 5

(12 weeks) timepoint according to RECIST version 1.1 [21].

In our patient cohort, four patients responded to therapy (one

complete responder [CR] and three partial responders [PR]),

six patients had an intermediate response (stable disease [SD])

and 10 patients failed to respond to therapy (progressive dis-

ease [PD]). The focus of this study was to identify immune

correlates of response and we therefore chose to primarily

compare patients who responded (CR/PR) and who failed to

respond (PD) when stratifying by clinical response. Age, sex,

and ECOG performance status were comparable across re-

sponse groups (Fig. 1A).

PBMCs isolated from collected blood samples were analysed

using ow cytometry and we rst sought to determine whether

PD-1 blockade altered the frequencies of CD4 and CD8 T

cells, as well as naïve and memory T cell subsets regardless of

response. We observed no differences in the CD4:CD8 T cell

ratio after treatment with pembrolizumab (Fig. 1B). Similarly,

frequencies of naïve and memory populations within CD4 and

CD8 T cells remained unchanged following PD-1 blockade

(Fig. 1C, Supplementary Fig. 1).

BL 6WK

12WK

2 mg/kg pembrolizumab every 3 weeks

Pre-cycle 5

Assessment

Sample collection

CR /

SD PD

age (s.d.)

70.5

(10.9)

75.0

(12.6)

72.4

(11.9)

%Male 25% 16.7% 50%

Median

ECOG PS

000

Figure 1. PASIP study cohort. Patients with advanced malignant melanoma were treated with 2 mg/kg pembrolizumab (αPD-1) every 3 weeks. Blood

samples were analysed using ﬂow cytometry at baseline and 6 and 12 weeks following treatment initiation. (A) Left: Graphic representation of PASIP

study protocol. BL = Baseline, 6WK = 6 weeks, 12WK = 12 weeks. Right: Patient age, sex, and ECOG performance status (PS) across response

groups. CR = complete response, PR = partial response, SD = stable disease, PD = progressive disease; s.d., standard deviation; PS, performance

status. (B) Ratio of CD4:CD8 T cells at indicated time points. Shown are means + SD. Two-tailed Student’s t test; ns, not signiﬁcant. (C) Distribution of

naïve (CD45RA+ CD62L+), central memory (CM, CD45RA- CD62L+), effector memory (EM, CD45RA- CD62L-) and terminally differentiated (TEMRA,

CD45RA+ CD62L-) CD4 (left) and CD8 (right) T cells. Shown are means of samples at indicated time points. BL, n = 19; 6WK, n = 20; 12WK, n = 16.

Downloaded from https://academic.oup.com/immunotherapyadv/article/3/1/ltad001/6972738 by Queen Mary University of London user on 01 March 2023

4 N.M. Edner et al.

Pembrolizumab precludes binding of the αPD-1 antibody

clone J105 used for ow cytometric staining [22], which

allowed us to examine occupancy of PD-1 by the thera-

peutic antibody on peripheral T cells. As expected, PD-1 on

CD4 and CD8 T cells was almost undetectable following

pembrolizumab treatment (Fig. 2A and B, Supplementary Fig.

2A). Surprisingly, when we split patients by clinical response,

we saw that more PD-1 tended to be detectable on CD4 T

cells in patients not responding to therapy, and this reached

statistical signicance at the 6-week timepoint (Fig. 2C,

Supplementary Fig. 2B). This was not the case for CD8 T cells

(Supplementary Fig. 2C). The higher residual PD-1 staining in

non-responders did not simply reect higher expression prior

to treatment, since patients with the highest expression at 6

Figure 2. More PD-1 detectable on CD4 T cells in individuals failing to respond to PD-1 blockade. PBMCs were stained with αPD-1 antibodies before

and after pembrolizumab treatment. (A) Representative ﬂow cytometry plots for PD-1 and CD45RA staining in CD4 T cells in one CR (top) and one

PD (bottom) patient at indicated time points. (B) PD-1+ frequency in CD4 T cells. BL, n = 19; 6WK, n = 20; 12WK, n = 16. (C) PD-1+ frequency in CD4

T cells in responders (CR/PR) and non-responders (PD). CR/PR, n = 4 (all time points); PD, n = 9 (BL), n = 10 (6WK), n = 6 (12WK). (B/C) Shown are

means + SD. Two-tailed Student’s t test; ****, P < 0.0001; *, P < 0.05; ns, not signiﬁcant.

Downloaded from https://academic.oup.com/immunotherapyadv/article/3/1/ltad001/6972738 by Queen Mary University of London user on 01 March 2023

Immune correlates of αPD-1 response in melanoma 5

weeks were not necessarily those with the highest expression

pre-treatment (Supplementary Fig. 2D). This suggests the pos-

sibility that in non-responders, PD-1 blockade on CD4 T cells

may be incomplete for reasons that remain unclear.

PD-1 blockade induces divergent immune changes

in responders and non-responders

PD-1 engagement suppresses T cell responses by dampening

signals necessary for efcient T cell stimulation [23].

Accordingly, blockade of PD-1 should enhance T cell activa-

tion and proliferation. We therefore evaluated expression of the

proliferation marker Ki67 in T cells following pembrolizumab

treatment. As expected, Ki67 expression was signicantly

increased in CD8 T cells 6 weeks following pembrolizumab

treatment initiation (Fig. 3A). There was also a trend towards

increased proliferation in CD4 T cells, however, this was not

statistically signicant (Supplementary Fig. 3A). To further

characterise the proliferating CD8 T cells, expression levels of

several markers were compared between the Ki67+ and Ki67-

fractions. Ki67+ CD8 T cells expressed higher intracellular

CTLA-4 and lower CD45RA and CD127 levels than the non-

proliferating CD8 T cells (Fig. 3B). Moreover, CD28 expres-

sion was higher on Ki67+ CD8 T cells when compared with

Ki67- CD8 T cells. While PD-1 can inhibit T cell responses

by suppressing TCR signalling [24, 25], there is also evidence

that PD-1 engagement can lead to the dephosphorylation

of CD28, thereby dampening CD28 downstream signalling

[26–28]. Stratication of CD28+Ki67+ and CD28-Ki67+

CD8 T cells by clinical response revealed that there was a

signicantly higher frequency of CD28+Ki67+ CD8 T cells in

responders than in non-responders at the 6-week timepoint,

with the CR patient showing the highest levels of Ki67 (Fig.

3C–E, Supplementary Fig. 3B). While a trend for increased

Ki67 expression in responders at the 6-week timepoint was

also observable when looking at the whole CD8 T cell popula-

tion this did not reach statistical signicance (Supplementary

Fig. 3C). Ki67 expression in CD4 T cells also did not differ

between clinical response groups (Supplementary Fig. 3D).

Taken together, these data suggest that response following

PD-1 blockade is marked by increased proliferation in CD28-

expressing CD8 T cells.

We were next interested in identifying changes occurring

specically in non-responders to pembrolizumab treatment.

When investigating CD4 T cell subsets, we found that in

patients that did not respond to therapy, the frequency of

regulatory T cells (Treg, CD25+CD127-Foxp3+CTLA-4+)

transiently increased 6 weeks post-treatment initiation and

then returned to baseline by the 12-week timepoint (Fig.

4A, Supplementary Fig. 4A). Interestingly, when evaluating

expression of Treg markers in non-responder samples using

median uorescence intensity (MFI) in all CD4 T cells, we

saw that only expression of CTLA-4 increased signicantly

following pembrolizumab treatment, while other markers,

including Foxp3, CD25, and CD127 remained consistent

over the treatment period (Fig. 4B). This increase was also

observed when analysing gated Treg (Supplementary Fig.

4B and C). Additionally, Treg in non-responders but not

responders showed increased expression of CXCR3 and ICOS

after therapy (Fig. 4C, Supplementary Fig. 4D) and there was

also a trend towards increased Ki67 expression in the Treg

compartment (Supplementary Fig. 4E). Therefore, an increase

in CTLA-4 expressing Treg is associated with non-response

following PD-1 blockade.

As frequencies of immune cell populations can have limited

applicability when clinically evaluating response to immuno-

therapy in individual patients, we sought to evaluate whether

assessing ratios of certain immune cell subsets would provide

a more robust approach for stratifying patient responses.

Since we saw increases in activated and proliferating Treg

(CXCR3+ICOS+ and Ki67+, respectively) in non-responders

following PD-1 blockade and conversely higher frequencies

of corresponding CD8 T cells in responders (Fig. 3;

Supplementary Fig. 5A), we investigated whether Treg:CD8

ratios would provide a way to distinguish clinical response.

Both the CXCR3+ICOS+ Treg to CXCR3+ICOS+ CD8 T

cell ratio and Ki67+ Treg to Ki67+CD28+ CD8 T cell ratio

were skewed to being signicantly higher in non-responders

6 weeks following pembrolizumab treatment initiation

(Fig. 4D and E, Supplementary Fig. 5B and C). Specically,

an average ratio of 1 CXCR3+ICOS+ CD8 T cell to 19.8

CXCR3+ICOS+ Treg was observed in non-responders while

responders showed an average ratio of 1 to 7.3. The ratio

of Ki67+CD28+ CD8 T cells to Ki67+ Treg was 1 to 14.9

in non-responders and 1 to 2.7 in responders. Importantly,

this does not simply reect the overall Treg:CD8 ratio which

was not signicantly different between responders and non-

responders at all timepoints investigated (Supplementary

Fig. 5D). Therefore, assessing the ratios of activated and

proliferating Treg and CD8 T cells may provide a way to

stratify clinical response following pembrolizumab treat-

ment. Additionally, we found that CXCR3+ICOS+ Treg and

CD8 T cells positively correlated with their Ki67 expressing

counterparts at the 6-week timepoint (Supplementary Fig.

5E and F), suggesting that CXCR3 and ICOS co-expression

could potentially be used as a surrogate marker for prolifera-

tion in these cell types.

Clinical response following PD-1 blockade can be

distinguished at baseline

Results presented thus far focus on response-specic changes

occurring in peripheral T cells following PD-1 blockade,

particularly 6 weeks post-treatment initiation. However, in

many malignancies, the therapeutic window is limited and

biomarkers to predict patient response to immunotherapy

before treatment would be highly benecial. Accordingly,

we were interested to see whether we could identify immune

signatures in baseline bleeds that could stratify patients based

on their clinical response following PD-1 blockade.

In order to maximise the information obtained from the

ow cytometry data, we decided to use the unsupervised

clustering algorithm FlowSOM [29] to identify as many im-

mune cell subsets as possible. Across the four ow cytometry

panels used for analysis, a total of 156 CD4 and CD8 T cell

clusters were detected (Supplementary Fig. 6). Following

pairwise correlation comparison, four of these clusters were

found to be highly correlated (Pearson correlation coefcient

> 0.95) to other clusters and therefore removed. This left 152

clusters which were used to conduct principal component

analysis (PCA). We reasoned that patient age may also inu-

ence clinical response and thus included this as a feature in

the analysis as well.

Remarkably, baseline bleeds of responders and non-

responders showed a clear separation along the rst prin-

cipal component (PC1), which accounts for 19.2% of the

variance in this dataset (Fig. 5A). We investigated the top 15

clusters contributing to PC1 in either direction to identify

Downloaded from https://academic.oup.com/immunotherapyadv/article/3/1/ltad001/6972738 by Queen Mary University of London user on 01 March 2023

6 N.M. Edner et al.

Figure 3. CD28-dependent increase in proliferating CD8 T cells indicative of response following PD-1 blockade. (A) Ki67+ frequency in CD8 T cells.

BL, n = 19; 6WK, n = 20; 12WK, n = 16. (B) MFI of indicated markers in Ki67+ or Ki67- CD8 T cells at the 6-week time point. n = 20. (C) Frequency of

Ki67+ CD28+ (top) and Ki67+ CD28- (bottom) in CD8 T cells stratiﬁed by response. CR/PR, n = 4 (all time points); SD, n = 6 (all time points); PD, n =

9 (BL), n = 10 (6WK), n = 6 (12WK). (D) tSNE projection of downsampled and pooled CD8 T cells of CR, PR, and PD samples. Colour indicates scaled

Ki67 expression. (E) tSNE projection shown in (D) of one CR and one PD sample at indicated time points. Colour indicates scaled Ki67 expression.

(A/C) Shown are means + SD. (B) Shown are box plots, with black horizontal line denoting median value, while box represents the IQRs (IQR, Q1–Q3

percentile) and whiskers show the minimum (Q1 − 1.5× IQR) and maximum (Q3 + 1.5× IQR) values. (A/B/C) Two-tailed Student’s t test; ****, P < 0.0001;

***, P < 0.001; **, P < 0.01; *, P < 0.05; ns, not signiﬁcant.

Downloaded from https://academic.oup.com/immunotherapyadv/article/3/1/ltad001/6972738 by Queen Mary University of London user on 01 March 2023

Immune correlates of αPD-1 response in melanoma 7

characteristics of the T cell populations driving this separa-

tion (Fig. 5B–D). One of the populations contributing most

to PC1 comprised CD8 T cells lacking expression of CD28

(panel2_CD8_9). Indeed, after manually gating CD28 expres-

sion, we saw that on average more than 50% of CD8 T cells in

patients that failed to respond were CD28- at baseline, while

this was only the case for about 25% of CD8 T cells in the

patients who responded to therapy (Fig. 5E, Supplementary

Figs. 3B and 7A). Conversely, one of the clusters associated

with response showed high Ki67 expression in CD28+ CD8

T cells (panel2_CD8_13). It is important to note that this

cluster was highly correlated with a Ki67 expressing CD4

conventional T cell cluster (panel2_CD4_16, Supplementary

Fig. 7B and C) which was consequently removed from the

features used for PCA. This, therefore, indicates that patients

responding to PD-1 blockade show higher baseline prolifera-

tion in CD8 as well as CD4 T cells.

Furthermore, we noted that clusters associated with non-

response to therapy predominantly exhibited an effector

memory or terminally differentiated effector memory phe-

notype, while clusters linked to response tended to be

central memory or naïve T cells. Among the latter, a

CD45RA+CD62L+ CD8 T cell cluster (panel4_CD8_14)

caught our attention, since it exhibited high CD38 expres-

sion and was associated with response following therapy.

We could conrm by manual gating that CD38hiCD62L+

CD8 T cells were indeed enriched in responders at baseline

(Supplementary Fig. 7D). Of possible interest, this population

also tended to be higher in frequency in responders compared

with non-responders at the 6- and 12-week timepoints and

displayed a remarkable increase in the patient showing a

complete response at the 6-week timepoint (Supplementary

Fig. 7D and E). Finally, two CD8 T cell clusters associated

with non-response following pembrolizumab treatment

showed high expression of the inhibitory receptors 2B4 and

KLRG1 (panel3_CD8_1 and panel3_CD8_4). Indeed, the

MFI of these two receptors in CD8 T cells was signicantly

higher in non-responders when compared with responders at

Figure 4. Ratio of activated and proliferating Treg and CD8 T cells is skewed in non-responders. (A) Frequency of Treg (CD25+ CD127- Foxp3+ CTLA-4+)

in CD4 T cells in responders and non-responders. (B) MFI of CD25, CD127, Foxp3, and intracellular CTLA-4 in CD4 T cells of non-responders (PD). (C)

CXCR3+ ICOS+ frequency in Treg (CD25+ CD127-). (D) Ratio of CXCR3+ ICOS+ frequency in Treg to CXCR3+ ICOS+ frequency in CD8 T cells. (E) Ratio

of Ki67+ frequency in Treg to Ki67+ CD28+ frequency in CD8 T cells. (A/C) Shown are means + SD. (B/D/E) Shown are box plots, with black horizontal

line denoting median value, while box represents the IQRs (IQR, Q1–Q3 percentile) and whiskers show the minimum (Q1 − 1.5× IQR) and maximum

(Q3 + 1.5× IQR) values. CR/PR, n = 4 (all time points); PD, n = 9 (BL), n = 10 (6WK), n = 6 (12WK). Two-tailed Student’s t test; *, P < 0.05; ns, not

signiﬁcant.

Downloaded from https://academic.oup.com/immunotherapyadv/article/3/1/ltad001/6972738 by Queen Mary University of London user on 01 March 2023

8 N.M. Edner et al.

Figure 5. Clinical response following PD-1 blockade can be distinguished at baseline. FlowSOM clustering was applied to CD4 and CD8 T cells stained

with four distinct ﬂow cytometry panels. (A) PCA on FlowSOM clusters of baseline samples of responders and non-responders. CR, n = 1; PR, n = 3;

PD, n = 7. (B) Top and bottom 15 FlowSOM clusters contributing to PC1 ordered by PC weight. (C) Heatmaps showing scaled MFI of indicated markers

in CD8 T cell clusters shown in (B). (D) Heatmaps showing scaled MFIs of indicated markers in CD4 T cell clusters shown in (B). Arrows in (C) and (D)

indicate directionality of associated PC1 weight. (E) CD28- frequency in CD8 T cells in baseline samples of responders and non-responders. (F) MFI

of 2B4 (top) and KLRG1 (bottom) in CD8 T cells normalised to MFI in naïve (CD45RA+ CCR7+) CD8 T cells. Data shown is from baseline samples of

responders and non-responders. (E/F) Shown are means + SD. CR/PR, n = 4; PD, n = 9. Two-tailed Student’s t test; **, P < 0.01; *, P < 0.05.

Downloaded from https://academic.oup.com/immunotherapyadv/article/3/1/ltad001/6972738 by Queen Mary University of London user on 01 March 2023

Immune correlates of αPD-1 response in melanoma 9

baseline (Supplementary Fig. 8A and B, Supplementary Fig.

9). Since the MFI is subjective to the uorochrome and ow

cytometer used, we opted to normalise using the MFI of 2B4

and KLRG1 on naïve CD8 T cells, where it was equivalent

between responders and non-responders (Supplementary Fig.

8C). Using this approach, we saw a 2.5× increase in 2B4 ex-

pression and 2× increase in KLRG1 expression in all CD8

T cells compared with naïve CD8 T cells of non-responders,

while there was only a 1.4× and 1.3× increase, respectively, in

the responders (Fig. 5F, Supplementary Fig. 8D).

Taken together, these results suggest that it may be pos-

sible to gain insights into the clinical response following

pembrolizumab treatment by utilising ow cytometric anal-

ysis of CD4 and CD8 T cell subsets at baseline. Specically,

higher baseline proliferation in CD4 and CD8 T cells is as-

sociated with a response to therapy, while non-response is

associated with a CD8 T cell compartment featuring more

CD28-negative and inhibitory receptor-expressing cells.

Discussion

While checkpoint inhibitor therapies are now an estab-

lished treatment option in cancer therapy, clinical response

of patients is variable. Robust biomarkers of response fol-

lowing PD-1 blockade are still lacking and peripheral

blood immunophenotyping remains incompletely explored.

Using T cell-focussed ow cytometric analysis we have

identied a number of features that correlate with response

to pembrolizumab in patients with advanced malignant

melanoma.

We found that patients that responded to PD-1 blockade ex-

perienced a greater increase in expression of the proliferation

marker Ki67 particularly in CD28-expressing CD8 T cells.

This is in accordance with previously published data showing

an increase in CD8 T cell proliferation that correlates with

clinical response and is largely driven by CD28-expressing

CD8 T cells [13, 27]. Furthermore, we show that the fre-

quency of proliferating CD28+ CD8 T cells directly correlates

with the frequency of CXCR3- and ICOS-expressing CD8 T

cells at the 6-week timepoint, suggesting that the latter could

be used as surrogate markers for PD-1 induced proliferation,

avoiding the need for intracellular Ki67 staining. ICOS ex-

pression on CD8 T cells responding to PD-1 blockade has

previously been shown [13], while work by Chow et al.

demonstrated that CXCR3 expression is important for αPD-1

efcacy in a mouse tumour model [30].

Conversely, patients that failed to respond to PD-1 blockade

showed a transient increase in Treg 6 weeks after treatment

initiation. It has previously been shown that more Treg and

a higher frequency of proliferating Treg can be found in pe-

ripheral blood of melanoma patients when compared with

healthy controls [12, 31, 32] and presence of Treg within

the tumour is associated with a worse prognosis [33], con-

sistent with the idea that Treg suppress anti-tumour immune

responses [34]. The response of CD8 T cells to PD-1 blockade

has been extensively studied, however evidence of modulation

of Treg has only recently started to emerge. While initial data

did not show any correlation between increases in Treg and

clinical outcome following PD-1 blockade [12], later reports

showed that presence of proliferating Treg within the tumour

as well as peripheral blood following PD-1 blockade was as-

sociated with a poor prognosis [35, 36], consistent with our

ndings. The difference in observations may be attributed to

the markers used to identify Treg in these studies, with our

data showing that it is particularly Treg expressing CTLA-4

that are increased following therapy in non-responders. This

upregulation of CTLA-4 on Treg could be a result of enhanced

activation following PD-1 blockade and higher levels of

CTLA-4 are associated with increased Treg function [37].

CTLA-4 expression is also a key feature of highly suppres-

sive Treg found to be enriched in tumour tissue of breast and

lung cancer patients, which are associated with more aggres-

sive disease [38–40]. Of note, recent analysis has highlighted

the potential for CTLA-4-expressing tumour-inltrating fol-

licular regulatory T cells (Tfr) to limit the effectiveness of

PD-1 inhibitors [41]. Together, these studies provide strong

biological context to our nding that non-responders are

characterised by increases in CTLA-4+ Treg. Since CTLA-4

is a key mediator of Treg suppression [42], it is possible that

inhibition of Treg function by anti-CTLA-4 antibodies might

prevent expanded Treg from limiting anti-tumour immu-

nity in these individuals, perhaps explaining why therapies

targeting both PD-1 and CTLA-4 work well together.

In our pre-treatment analysis, we identied that responders

showed a higher frequency of proliferating CD28-expressing

CD8 T cells as well as proliferating CD4 conventional T cells,

suggestive of ongoing immunity, potentially encompassing

anti-tumour immune responses. Conversely, non-responders

had higher frequencies of CD28- CD8 T cells, that are less

responsive to PD-1 therapy, and showed higher expression

of inhibitory receptors 2B4 and KLRG1 at baseline. CD28

expression on CD8 T cells is known to be downregulated by

repeated antigen stimulation and CD28- CD8 T cells accumu-

late with age [43, 44]. Both 2B4 and KLRG1 have previously

been linked to CD8 T cell exhaustion in chronic viral infec-

tion [45, 46] and targeting KLRG1 in combination with PD-1

blockade has been evaluated in an in vivo cancer model [47].

Interestingly, recent work has identied a population of CD8

T cells with immunoregulatory function that have lost CD28

and express both 2B4 and KLRG1 [48]. It is possible that the

non-responders in our cohort had higher levels of these sup-

pressive CD8 T cells prior to therapy.

While complete responses following PD-1 pathway

blockade have been observed, they only occur in a limited

number of individuals. Biomarkers to identify this patient

group would therefore be highly benecial. In our patient

cohort, only one individual showed a complete response to

pembrolizumab, precluding us from further investigating

correlates of response. However, we did observe a notable

increase in CD38hiCD62L+ CD8 T cells 6 weeks after treat-

ment initiation in this patient. This cell subset bears some re-

semblance to the circulating exhausted CD8 T cells shown

by Huang et al. to be reinvigorated following PD-1 blockade

[12] and could conceivably demark patients responding ex-

ceptionally well to therapy.

Finally, our anti-PD-1 antibody staining revealed a higher

frequency of unoccupied PD-1 on CD4 T cells at the 6-week

timepoint in non-responders. Most reports evaluate PD-1

expression using antibodies targeting the Fc region of the

relevant therapeutic antibody and there is relatively little

data evaluating PD-1 occupancy in patients following treat-

ment. Data from Das et al. showed that even when PD-1

is entirely occupied by therapeutic antibody in peripheral

blood, blockade of PD-1 on TILs is still incomplete [49].

Downloaded from https://academic.oup.com/immunotherapyadv/article/3/1/ltad001/6972738 by Queen Mary University of London user on 01 March 2023

10 N.M. Edner et al.

This analysis could suggest that if PD-1 blockade is not com-

plete in the periphery, blockade within the tumour may not

be sufcient either, potentially affecting treatment efcacy.

Additionally, Zappasodi et al. made use of an αPD-1 anti-

body that is not blocked by therapeutic αPD-1 antibodies

to show that PD-1hi CD4 T cells are decreased following

PD-1 blockade and that this reduction was less prevalent

in patients that failed to respond to therapy [50]. While

we cannot directly compare these PD-1hi CD4 T cells with

the PD-1+ CD4 T cells we observe after therapy in non-

responders, our data suggest that the relationship between

treatment efcacy and PD-1 receptor occupancy warrants

further investigation.

In summary, by conducting peripheral blood

immunophenotyping in patients with advanced malig-

nant melanoma treated with pembrolizumab, we identied

skewing of the peripheral immune response both at baseline

and shortly after treatment initiation that correlated with re-

sponse following PD-1 blockade. Specically, patients that

failed to respond to therapy displayed a more immune sup-

pressive phenotype with low CD28 expression on CD8 T

cells and an increase in Treg following therapy. Additionally,

we found that ratios of activated and proliferating Treg and

CD8 T cells could stratify patients by clinical response. Our

results were obtained in a limited number of patients, and it

will be important to validate these ndings in larger patient

cohorts. Nonetheless, our work suggests that T cell-directed

ow cytometric assays could provide a further tool to inform

rapid treatment decisions.

Supplementary material

Supplementary data are available at Immunotherapy

Advances online.

Figure S1. Frequencies of naïve and memory CD4 and

CD8 T cells does not change following PD-1 blockade. (A)

Frequency of naïve (top left, CD45RA+ CD62L+), central

memory (CM, top right, CD45RA- CD62L+), effector mem-

ory (EM, bottom left, CD45RA- CD62L-) and terminally

differentiated (TEMRA, bottom right, CD45RA+ CD62L-)

in CD4 T cells. (B) Frequency of naïve (top left, CD45RA+

CD62L+), central memory (CM, top right, CD45RA-

CD62L+), effector memory (EM, bottom left, CD45RA-

CD62L-) and terminally differentiated (TEMRA, bottom

right, CD45RA+ CD62L-) in CD8 T cells. Shown are means

+ s.d.. BL, n = 19; 6WK, n = 20; 12WK, n = 16. Two-tailed

Student’s t test; ns, not signicant.

Figure S2. PD-1 detection in CD8 T cells following PD-1

blockade is the same in responders and non-responders. (A)

PD-1+ frequency in CD8 T cells. BL, n = 19; 6WK, n = 20;

12WK, n = 16. (B) PD-1+ frequency in CD4 T cells strati-

ed by response. (C) PD-1+ frequency in CD8 T cells strat-

ied by response. (D) (top) PD-1+ frequency in CD4 T cells

in CR, PR, SD and PD patients. Points from same patient are

connected by lines. Colour indicates PD-1+ frequency at base-

line. (bottom) Ranking of PD-1 expression in CR, PR, SD and

PD patients at baseline and the 6-week timepoint. For P23 no

baseline bleed was available. (B/C/D) CR/PR, n = 4 (all time

points); SD, n = 6 (all time points); PD, n = 9 (BL), n = 10

(6WK), n = 6 (12WK). Shown are means + s.d.. Two-tailed

Student’s t test; ****, P < 0.0001; *, P < 0.05; ns, not signif-

icant.

Figure S3. Increase in proliferation in CD4 T cells does

not distinguish responders and non-responders. (A) Ki67+

frequency in CD4 T cells. (B) Representative ow cytometry

plots showing Ki67 and CD28 expression in CD8 T cells

in baseline, 6-week and 12-week bleeds of one PR and one

PD patient. (C) Ki67+ frequency in CD8 T cells stratied by

response. (D) Frequency of Ki67+ (left), Ki67+ CD28+ (mid-

dle) and Ki67+ CD28- (right) in CD4 T cells stratied by re-

sponse. (A) BL, n = 19; 6WK, n = 20; 12WK, n = 16. (C/D)

CR/PR, n = 4 (all time points); SD, n = 6 (all time points); PD,

n = 9 (BL), n = 10 (6WK), n = 6 (12WK). Shown are means +

s.d.. Two-tailed Student’s t test; ns, not signicant.

Figure S4. CTLA-4+ Treg are transiently increased in

non-responders following PD-1 blockade. (A) Frequency of

Treg (CD25+ CD127- Foxp3+ CTLA-4+) in CD4 T cells.

(B) MFI of intracellular CTLA-4 in Treg (CD25+ CD127-)

of non-responders. (C) Scaled histogram showing intra-

cellular CTLA-4 expression in naïve T cells (lled) or Treg

(open) from a non-responder at the indicated time points.

(D) CXCR3+ ICOS+ frequency in Treg (CD25+ CD127-). (E)

Ki67+ frequency in Treg. (A/D/E) Shown are means + s.d..

(B) Shown are box plots, with black horizontal line denoting

median value, while box represents the IQRs (IQR, Q1–Q3

percentile) and whiskers show the minimum (Q1 − 1.5× IQR)

and maximum (Q3 + 1.5× IQR) values. (A/B/D/E) CR/PR, n =

4 (all time points); SD, n = 6 (all time points); PD, n = 9 (BL),

n = 10 (6WK), n = 6 (12WK). Two-tailed Student’s t test; *, P

< 0.05; ns, not signicant.

Figure S5. CXCR3 and ICOS expression correlates with

proliferation of Treg and CD8 T cells. (A) CXCR3+ ICOS+

frequency in CD8 T cells. (B) Ratio of CXCR3+ ICOS+ fre-

quency in Treg to CXCR3+ ICOS+ frequency in CD8 T cells.

quency in CD8 T cells. (D) Ratio of Treg frequency (CD25+

CD127- Foxp3+ CTLA-4+) to CD8 T cell frequency. (E)

Pearson correlation of Ki67+ frequency in Treg (CD25+

CD127- Foxp3+ CTLA-4+) to CXCR3+ ICOS+ frequency

in Treg (CD25+ CD127-). (F) Pearson correlation of Ki67+

frequency to CXCR3+ ICOS+ frequency in CD8 T cells.

(A) Shown are means + s.d.. (B/C/D) Shown are box plots,

with black horizontal line denoting median value, while box

represents the IQRs (IQR, Q1–Q3 percentile) and whiskers

show the minimum (Q1 − 1.5× IQR) and maximum (Q3 +

1.5× IQR) values. (A/B/C/D) CR/PR, n = 4 (all time points);

SD, n = 6 (all time points); PD, n = 9 (BL), n = 10 (6WK), n =

6 (12WK). Two-tailed Student’s t test; ***, P < 0.001; *, P <

0.05; ns, not signicant. (E/F) CR, n = 1; PR, n = 3; SD, n = 6;

PD, n = 10. Pearson’s R and associated p value are depicted on

plots. Black line only for visualisation purposes.

Figure S6. Heatmaps of maker expression in FlowSOM

clusters. FlowSOM clustering was applied to CD4 and CD8 T

cells stained with four distinct ow cytometry panels. Shown

are heatmaps of scaled MFIs of indicated markers in CD4

(left) and CD8 (right) T cells.

Figure S7. Clinical response following PD-1 blockade can

be distinguished using baseline bleeds. (A) CD28- frequency

in CD8 T cells in baseline samples. (B) Pearson correlation

of frequencies of FlowSOM clusters panel2_CD4_16 and

panel2_CD8_13. (C) Pearson correlation of manually gated

Ki67+ frequency in Tconv (non-Treg) and Ki67+ CD28+

frequency in CD8 T cells. (D) CD62L+ CD38hi frequency

in CD8 T cells. (E) Flow cytometry plots showing CD62L

Downloaded from https://academic.oup.com/immunotherapyadv/article/3/1/ltad001/6972738 by Queen Mary University of London user on 01 March 2023

Immune correlates of αPD-1 response in melanoma 11

and CD38 expression in CD8 T cells in baseline and 6-week

bleeds of CR patient. (A/D) Shown are means + s.d.. CR/PR,

n = 4 (all time points); SD, n = 6 (all time points); PD, n =

9 (BL), n = 10 (6WK), n = 6 (12WK). Two-tailed Student’s t

test; **, P < 0.01; *, P < 0.05; ns, not signicant. (B/C) CR,

n = 1; PR, n = 3; SD, n = 6; PD, n = 7. Pearson’s R and as-

sociated p value are depicted on plots. Black line only for

visualisation purposes.

Figure S8. CD8 T cells of non-responders have higher ex-

pression of 2B4 and KLRG1 at baseline. (A) MFI of 2B4

(left) and KLRG1 (right) in CD8 T cells in baseline samples.

(B) Representative ow cytometry plots showing 2B4 and

KLRG1 expression in CD8 T cells in baseline bleed of one PR

patient and one PD patient. (C) MFI of 2B4 (left) and KLRG1

(right) in naïve (CD45RA+ CCR7+) CD8 T cells in baseline

samples. (D) MFI of 2B4 (left) and KLRG1 (right) in CD8 T

cells normalised to MFI in naïve (CD45RA+ CCR7+) CD8 T

cells. Data shown is from baseline samples. (A/C/D) Shown

are means + s.d.. CR/PR, n = 4; SD, n = 6; PD, n = 9. Two-

tailed Student’s t test; **, P < 0.01; ns, not signicant.

Figure S9. Gating strategies. Shown are representative

gating strategies for relevant CD4 and CD8 T cell populations

in indicated ow cytometry panels. For all panels, cells in rst

gate are live, singlet Lymphocytes. Grey background indicates

same parent gate.

Table S1. Patient characteristics. Shown are characteristics

of patients enrolled in the PASIP study.

Table S2. Flow cytometry surface panels. Shown are the

four ow cytometry panels of antibodies that were used for

surface staining of PBMCs in this study.

Acknowledgements

We gratefully acknowledge the contributions of Sachuda

Veerasamy, Leah Meaden, Kirsty Prout and Rebecca Sargent

in organising blood sample provision. The Editor-in-Chief,

Tim Elliott, and handling editor, Stephanie Dougan, would

like to thank the following reviewers, an anonymous reviewer

and Andreas Beilhack, for their contribution to the publica-

tion of this article.

Author contributions

N.M.E. assisted with the experiments, analysed the data, pre-

pared the gures and wrote the manuscript. E.N. performed

the experiments and edited the manuscript. L.P., C.J.W., A.F.,

A.K., V.O., E.M.R., F.H., Y.E. and L.P.H. assisted with the

experiments and edited the manuscript. T.T., A.S. and A.P.

provided clinical samples and edited the manuscript. D.C.

conceptualised and wrote the study protocol, applied for fund-

ing, supervised clinical sample provision, assessed treatment

response and edited the manuscript. L.S.K.W conceptualised

and supervised the study, designed the experiments, applied

for funding and wrote the manuscript.

Funding

Funding for this study was provided by a research grant

from the Investigator-Initiated Studies Program of Merck

Sharp & Dohme Limited. The opinions expressed in this pa-

per are those of the authors and do not necessarily repre-

sent those of Merck Sharp & Dohme Limited. This work as

also supported by CancerResearchUK grant no. C58264/

A26593 and Medical Research Council grant no. MR/

N001435/1. This project has received funding from the

European Union’s Horizon 2020 research and innovation

programme under the Marie Skłodowska-Curie grant agree-

ment no. 955321.

Conﬂict of interest

L.S.K.W. and D.C. received funding from Merck Sharp &

Dohme Limited for this study. T.T. has received travel grants

and honoraria from Merck Sharp & Dohme Limited.

Permission to reproduce

Not applicable to this manuscript.

Clinical trial registration

The PASIP research study is listed at ClinicalTrials.gov and

can be found under the identier NCT02909348. The proto-

col and consent document of this trial were approved by the

London - Bromley Research Ethics Committee (16/LO/1296).

ARRIVE guidelines

Not applicable to this manuscript.

Data availability

Deidentied individual participant data supporting the

ndings of this study are available from the corresponding

author upon request.

References

1. Robert C. A decade of immune-checkpoint inhibitors in cancer

therapy. Nat Commun 2020; 11(1):3801.https://doi.org/10.1038/

s41467-020-17670-y

2. Hamid O, Robert C, Daud A et al. Five-year survival outcomes

for patients with advanced melanoma treated with pembrolizumab

in KEYNOTE-001. Ann Oncol 2019; 30(4):582–8. https://doi.

org/10.1093/annonc/mdz011

3. Ribas A, Wolchok JD. Cancer immunotherapy using check-

point blockade. Science 2018; 359(6382):1350–5. https://doi.

org/10.1126/science.aar4060

4. Daud AI, Wolchok JD, Robert C et al. Programmed death-ligand 1

expression and response to the anti-programmed death 1 antibody

pembrolizumab in melanoma. J Clin Oncol 2016; 34(34):4102–9.

https://doi.org/10.1200/jco.2016.67.2477

5. Topalian SL, Hodi FS, Brahmer JR et al. Safety, activity, and

immune correlates of anti–PD-1 antibody in cancer. N Engl

J Med 2012; 366(26):2443–54. https://doi.org/10.1056/

nejmoa1200690

6. Gubin MM, Zhang X, Schuster H et al. Checkpoint blockade

cancer immunotherapy targets tumour-specic mutant antigens.

Nature 2014; 515(7528):577–81. https://doi.org/10.1038/na-

ture13988

7. Rizvi NA, Hellmann MD, Snyder A et al. Mutational landscape

determines sensitivity to PD-1 blockade in non-small cell lung

cancer. Science 2015; 348(6230):124–8. https://doi.org/10.1126/

science.aaa1348

8. Daud AI, Loo K, Pauli ML et al. Tumor immune proling predicts

response to anti-PD-1 therapy in human melanoma. J Clin Invest

2016; 126(9):3447–52. https://doi.org/10.1172/jci87324

Downloaded from https://academic.oup.com/immunotherapyadv/article/3/1/ltad001/6972738 by Queen Mary University of London user on 01 March 2023

12 N.M. Edner et al.

9. Tumeh PC, Harview CL, Yearley JH et al. PD-1 blockade induces

responses by inhibiting adaptive immune resistance. Nature 2014;

515(7528):568–71. https://doi.org/10.1038/nature13954

10. Gros A, Parkhurst MR, Tran E et al. Prospective identication

of neoantigen-specic lymphocytes in the peripheral blood of

melanoma patients. Nat Med 2016; 22(4):433–8. https://doi.

org/10.1038/nm.4051

11. Gros A, Tran E, Parkhurst MR et al. Recognition of human gas-

trointestinal cancer neoantigens by circulating PD-1+ lymphocytes.

J Clin Invest 2019; 129(11):4992–5004. https://doi.org/10.1172/

jci127967

12. Huang AC, Postow MA, Orlowski RJ et al. T-cell invigoration to

tumour burden ratio associated with anti-PD-1 response. Nature

2017; 545(7652):60–5. https://doi.org/10.1038/nature22079

13. Kamphorst AO, Pillai RN, Yang S et al. Proliferation of PD-1+

CD8 T cells in peripheral blood after PD-1-targeted therapy in lung

cancer patients. Proc Natl Acad Sci USA 2017; 114(19):4993–8.

https://doi.org/10.1073/pnas.1705327114

14. Mazzaschi G, Facchinetti F, Missale G et al. The circulating pool of

functionally competent NK and CD8+ cells predicts the outcome

of anti-PD1 treatment in advanced NSCLC. Lung Cancer 2019;

127:153–63. https://doi.org/10.1016/j.lungcan.2018.11.038

15. Kim KH, Cho J, Ku BM et al. The rst-week proliferative response

of peripheral blood PD-1+CD8+ T cells predicts the response

to Anti-PD-1 therapy in solid tumors. Clin Cancer Res 2019;

25(7):2144–54.https://doi.org/10.1158/1078-0432.ccr-18-1449

16. Juliá EP, Mandó P, Rizzo MM et al. Peripheral changes in immune

cell populations and soluble mediators after anti-PD-1 therapy

in non-small cell lung cancer and renal cell carcinoma patients.

Cancer Immunol Immunother 2019; 68(10):1585–96. https://doi.

org/10.1007/s00262-019-02391-z

17. Zuazo M, Arasanz H, Fernández-Hinojal G et al. Functional sys-

temic CD4 immunity is required for clinical responses to PD -L1/ PD

-1 blockade therapy. EMBO Mol Med 2019; 11(7):e10293.https://

doi.org/10.15252/emmm.201910293

18. Ferrara R, Naigeon M, Auclin E et al. Circulating T-cell

immunosenescence in patients with advanced non-small cell

lung cancer treated with single-agent PD-1/PD-L1 inhibitors or

platinum-based chemotherapy. Clin Cancer Res 2021; 27(2):492–

503. https://doi.org/10.1158/1078-0432.ccr-20-1420

19. Edner NM, Wang CJ, Petersone L et al. Predicting clinical response

to costimulation blockade in autoimmunity. Immunother Adv

2021; 1(1):ltaa003. https://doi.org/10.1093/immadv/ltaa003

20. Edner NM, Heuts F, Thomas N et al. Follicular helper T cell proles

predict response to costimulation blockade in type 1 diabetes. Nat

Immunol 2020; 21(10):1244–55. https://doi.org/10.1038/s41590-

020-0744-z

21. Eisenhauer EA, Therasse P, Bogaerts J et al. New response evalu-

ation criteria in solid tumours: revised RECIST guideline (version

1.1). Eur J Cancer 2009; 45(2):228–47.https://doi.org/10.1016/j.

ejca.2008.10.026

22. Zelba H, Bochem J, Pawelec G et al. Accurate quantication of

T-cells expressing PD-1 in patients on anti-PD-1 immunotherapy.

Cancer Immunol Immunother 2018; 67(12):1845–51. https://doi.

org/10.1007/s00262-018-2244-7

23. Sharpe AH, Pauken KE. The diverse functions of the PD1 inhibi-

tory pathway. Nat Rev Immunol 2018; 18(3):153–67. https://doi.

org/10.1038/nri.2017.108

24. Yokosuka T, Takamatsu M, Kobayashi-Imanishi W et al. Pro-

grammed cell death 1 forms negative costimulatory microclusters

that directly inhibit T cell receptor signaling by recruiting phos-

phatase SHP2. J Exp Med 2012; 209(6):1201–17. https://doi.

org/10.1084/jem.20112741

25. Mizuno R, Sugiura D, Shimizu K et al. PD-1 primarily targets TCR

signal in the inhibition of functional T cell activation. Front Immunol

2019; 10:630. https://doi.org/10.3389/mmu.2019.00630

26. Hui E, Cheung J, Zhu J et al. T cell costimulatory receptor

CD28 is a primary target for PD-1-mediated inhibition. Sci-

ence 2017;355(6332):1428–33. https://doi.org/10.1126/science.

aaf1292

27. Kamphorst AO, Wieland A, Nasti T et al. Rescue of exhausted

CD8 T cells by PD-1-targeted therapies is CD28-dependent. Sci-

ence 2017; 355(6332):1423–7. https://doi.org/10.1126/science.

aaf0683

28. Walker LSK. PD-1 and CTLA-4: two checkpoints, one pathway?

Sci Immunol 2017; 2(11):eaan3864. https://doi.org/10.1126/

sciimmunol.aan3864

29. Van Gassen S, Callebaut B, Van Helden MJ et al. FlowSOM:

using self-organizing maps for visualization and interpretation of

cytometry data. Cytom Part A 2015; 87(7):636–45. https://doi.

org/10.1002/cyto.a.22625

30. Chow MT, Ozga AJ, Servis RL et al. Intratumoral activity of the

CXCR3 chemokine system is required for the efcacy of anti-PD-1

therapy. Immunity 2019; 50(6):1498–1512.e5. https://doi.

org/10.1016/j.immuni.2019.04.010

31. Jandus C, Bioley G, Speiser DE et al. Selective accumulation of

differentiated FOXP3+ CD4+ T cells in metastatic tumor lesions

from melanoma patients compared to peripheral blood. Cancer

Immunol Immunother 2008; 57(12):1795–805. https://doi.

org/10.1007/s00262-008-0507-4

32. Krieg C, Nowicka M, Guglietta S et al. High-dimensional single-

cell analysis predicts response to anti-PD-1 immunotherapy. Nat

Med 2018; 24(2):144–53. https://doi.org/10.1038/nm.4466

33. Baumgartner JM, Gonzalez R, Lewis KD et al. Increased sur-

vival from stage IV melanoma associated with fewer regulatory

T cells. J Surg Res 2009; 154(1):13–20. https://doi.org/10.1016/j.

jss.2008.04.043

34. Antony PA, Piccirillo CA, Akpinarli A et al. CD8+ T cell immu-

nity against a tumor/self-antigen is augmented by CD4+ T helper

cells and hindered by naturally occurring T regulatory cells.

J Immunol 2005; 174(5):2591–601. https://doi.org/10.4049/

jimmunol.174.5.2591

35. Huang AC, Orlowski RJ, Xu X et al. A single dose of neoadjuvant

PD-1 blockade predicts clinical outcomes in resectable melanoma.

Nat Med 2019; 25(3):454–61. https://doi.org/10.1038/s41591-

019-0357-y

36. Kagamu H, Kitano S, Yamaguchi O et al. CD4+ T-cell immunity

in the peripheral blood correlates with response to Anti-PD-1

therapy. Cancer Immunol Res 2020; 8(3):334–44. https://doi.

org/10.1158/2326-6066.cir-19-0574

37. Ovcinnikovs V, Ross EM, Petersone L et al. CTLA-4-mediated

transendocytosis of costimulatory molecules primarily targets mi-

gratory dendritic cells. Sci Immunol 2019; 4(35):eaaw0902. https://

doi.org/10.1126/sciimmunol.aaw0902

38. Alvisi G, Brummelman J, Puccio S et al. IRF4 instructs effector

Treg differentiation and immune suppression in human cancer.

J Clin Invest 2020; 130(6):3137–50. https://doi.org/10.1172/

jci130426

39. Plitas G, Konopacki C, Wu K et al. Regulatory T cells exhibit dis-

tinct features in human breast cancer. Immunity 2016; 45(5):1122–

34. https://doi.org/10.1016/j.immuni.2016.10.032

40. De Simone M, Arrigoni A, Rossetti G et al. Transcriptional land-

scape of human tissue lymphocytes unveils uniqueness of tumor-

inltrating T regulatory cells. Immunity 2016; 45(5):1135–47.

https://doi.org/10.1016/j.immuni.2016.10.021

41. Eschweiler S, Clarke J, Ramírez-Suástegui C et al. Intratumoral fol-

licular regulatory T cells curtail anti-PD-1 treatment efcacy. Nat

Immunol 2021; 22(8):1052–63. https://doi.org/10.1038/s41590-

021-00958-6

42. Walker LSK. Treg and CTLA-4: two intertwining pathways to im-

mune tolerance. J Autoimmun 2013; 45(100):49–57. https://doi.

org/10.1016/j.jaut.2013.06.006

43. Posnett DN, Sinha R, Kabak S et al. Clonal populations of T cells

in normal elderly humans: the T cell equivalent to “Benign Mono-

clonal Gammapathy.”. J Exp Med 1994; 179(2):609–618. https://

doi.org/10.1084/jem.179.2.609

Downloaded from https://academic.oup.com/immunotherapyadv/article/3/1/ltad001/6972738 by Queen Mary University of London user on 01 March 2023

Immune correlates of αPD-1 response in melanoma 13

44. Valenzuela HF, Effros RB. Divergent telomerase and CD28 expres-

sion patterns in human CD4 and CD8 T cells following repeated

encounters with the same antigenic stimulus. Clin Immunol 2002;

105(2):117–25. https://doi.org/10.1006/clim.2002.5271

45. Blackburn SD, Shin H, Haining WN et al. Coregulation of CD8+

T cell exhaustion by multiple inhibitory receptors during chronic

viral infection. Nat Immunol 2009; 10(1):29–37. https://doi.

org/10.1038/ni.1679

46. Utzschneider DT, Charmoy M, Chennupati V et al. T cell factor

1-expressing memory-like CD8+ T cells sustain the immune re-

sponse to chronic viral infections. Immunity 2016; 45(2):415–27.

https://doi.org/10.1016/j.immuni.2016.07.021

47. Tata A, Dodard G, Fugère C et al. Combination blockade of KLRG1

and PD-1 promotes immune control of local and disseminated

cancers. Oncoimmunology 2021; 10(1):1933808. https://doi.org/

10.1080/2162402X.2021.1933808

48. Li J, Zaslavsky M, Su Y et al. KIR+CD8+ T cells suppress patho-

genic T cells and ar active in autoimmune diseases and COVID-19.

Science 2022; 376(6590):eabi9591. https://doi.org/10.1126/sci-

ence.abi9591

49. Das R, Verma R, Sznol M et al. Combination therapy with Anti–

CTLA-4 and anti–PD-1 leads to distinct immunologic changes

in vivo. J Immunol 2015; 194(3):950–9. https://doi.org/10.4049/

jimmunol.1401686

50. Zappasodi R, Budhu S, Hellmann MD et al. Non-conventional In-

hibitory CD4+Foxp3−PD-1hi T cells as a biomarker of immune

checkpoint blockade activity. Cancer Cell 2018; 33(6):1017–1032.

e7. https://doi.org/10.1016/j.ccell.2018.05.009

Downloaded from https://academic.oup.com/immunotherapyadv/article/3/1/ltad001/6972738 by Queen Mary University of London user on 01 March 2023

Do you think that responders are likely to have more antigen experienced PD1+ve stem cell like CD8 T cells, whilst non-responders have more exhausted PD1+ve CD8 T cells? Can you speculate on why you saw a difference in the frequency of PD1+ CD4 T cells between clinical response groups but not a concomitant increase in CD4 T cell proliferation? Do you think that these phenotypes would also been seen and predict response in other cancer patients or is it melanoma-speicific? Did you look at whether there was any correlation between the CD4:CD8 ratio at baseline and response to Pembrolizumab?

Comments

Products

Project