Early changes of peripheral circulating immune subsets induced by PD-1 inhibitors in patients with advanced malignant melanoma and non-small cell lung cancer

Abstract

Background

Immune checkpoint inhibitors (ICIs), including those targeting PD-1, are currently used in a wide range of tumors, but only 20–40% of patients achieve clinical benefit. The objective of our study was to find predictive peripheral blood-based biomarkers for ICI treatment.

Methods

In 41 patients with advanced malignant melanoma (MM) and NSCLC treated with PD-1 inhibitors, we analyzed peripheral blood-based immune subsets by flow cytometry before treatment initialization and the second therapy dose. Specifically, we assessed basic blood differential count, overall T cells and their subgroups, B cells, and myeloid-derived suppressor cells (MDSC). In detail, CD4 + and CD8 + T cells were assessed according to their subtypes, such as central memory T cells (TCM), effector memory T cells (TEM), and naïve T cells (TN). Furthermore, we also evaluated the predictive value of CD28 and ICOS/CD278 co-expression on T cells.

Results

Patients who achieved disease control on ICIs had a significantly lower baseline proportion of CD4 + TEM (p = 0.013) and tended to have a higher baseline proportion of CD4 + TCM (p = 0.059). ICI therapy-induced increase in Treg count (p = 0.012) and the proportion of CD4 + TN (p = 0.008) and CD28 + ICOS- T cells (p = 0.012) was associated with disease control. Patients with a high baseline proportion of CD4 + TCM and a low baseline proportion of CD4 + TEM showed significantly longer PFS (p = 0.011, HR 2.6 and p ˂ 0.001, HR 0.23, respectively) and longer OS (p = 0.002, HR 3.75 and p ˂ 0.001, HR 0.15, respectively). Before the second dose, the high proportion of CD28 + ICOS- T cells after ICI therapy initiation was significantly associated with prolonged PFS (p = 0.017, HR 2.51) and OS (p = 0.030, HR 2.69). Also, a high Treg count after 2 weeks of ICI treatment was associated with significantly prolonged PFS (p = 0.016, HR 2.33).

Conclusion

In summary, our findings suggest that CD4 + TEM and TCM baselines and an early increase in the Treg count induced by PD-1 inhibitors and the proportion of CD28 + ICOS- T cells may be useful in predicting the response in NSCLC and MM patients.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-024-13351-x.

Introduction

Immune checkpoint inhibitors (ICIs) have shown unprecedented efficacy in therapy of several solid tumor histotypes. Although in a subset of patients, PD-1 inhibitors transformed previously terminal diseases such as malignant melanoma (MM) into long-term survival (5-year overall survival rate of 44% in patients treated with nivolumab) [1] still in different tumor types, only 20–45% of patients respond to treatment [2]. Moreover, one of the biggest challenges associated with ICI treatment is the occurrence of immune-related adverse events (irAEs), as ICIs can not only boost the antitumor T cell immune response but also activate self-reactive T cells, resulting in attacking normal tissues or organs. These irAEs can occur in up to 90% of patients treated with an anti-cytotoxic T lymphocyte-associated protein-4 (CTLA-4) antibody [3] and up to 70% of patients treated with an anti-programmed cell death protein-1 (PD-1) antibody [4]. Thus, proper patient selection is crucial to prevent unnecessary treatment and the risk of irAE onset.

Currently, there are three Food and Drug Administration (FDA) approved tumor tissue-bound predictors for ICIs: expression of programmed death-ligand 1 (PD-L1) protein, microsatellite instability/defective mismatch repair (MSI/dMMR), and tumor mutational burden (TMB) [5]. However, each predictor has its limitations. To define PD-L1 and TMB, various assays specific for each tumor and ICI type are required. Also, thresholds differ according to tumor type, scoring systems, specific ICIs, and assay platforms. Although the MSI/dMMR is a powerful predictor for ICI response, its main limitation is prevalence among the tumor types. For example, the MSI/dMMR prevalence is the highest in colorectal cancer (CRC), which still represents only 17% of all CRCs [6]. Other tumor types where the ICI treatment is the most frequently used, such as non-small cell lung cancer (NSCLC), breast cancer, melanoma, and kidney cancer, have a low prevalence or no MSI/dMMR data are available. Nevertheless, the common limitation of these predictors is their dependence on tumor tissue availability, which is burdened by spatial and temporal heterogeneity. Therefore, there is a need to obtain a non-invasive, cost-effective peripheral blood-derived biomarker for prognostic and predictive utility in patients receiving ICI. Although, to this date, several studies have proposed different peripheral blood cell subsets associated with disease prognosis or ICI treatment efficacy, the prognostic significance of peripheral blood immune profiling remains limited [7].

Here, we aim to assess in advanced MM and NSCLC patients treated with PD-1 inhibitors comprehensive blood cell subsets in peripheral blood, such as white blood cell count and differential count, T cells and their subsets (cytotoxic, helper, regulatory, gamma-delta, natural killer-like T cells), B cells and myeloid-derived suppressor cells. According to the functional status, CD4 + and CD8 + T cells were classified into phenotypes: central memory T cells (CD45RO + CD27+) (TCM), effector memory T cells (CD45RO + CD27-) (TEM), naïve/early T cells (CD45RO-CD27+) (TN). Furthermore, based on preclinical data showing that costimulatory receptors such as CD28 are critical in PD-1 inhibition, we also analyzed the predictive value of CD28 and ICOS expression on T cells [8].

Materials and methods

Patients

In this study, we prospectively enrolled 41 patients with advanced or metastatic MM and NSCLC treated with monotherapy anti-PD-1 immune checkpoint inhibitor: Nivolumab (anti-PD-1 ICI, Opdivo^®), Pembrolizumab (anti-PD-1 ICI, Keytruda^®) from 2017 to 2021 at Masaryk Memorial Cancer Institute in Brno, Czech Republic. Patients received either nivolumab monotherapy with a dosage of 240 mg IV every 2 weeks or pembrolizumab monotherapy with 200 mg IV every 3 weeks. The study follow-up ended in March 2022. This study analyzed the peripheral circulating immune subsets before treatment (1st day of ICI treatment) as a baseline level and before the second dose (11th – 22nd day of ICI treatment). Inclusion criteria were age over 18, advanced/metastatic MM or NSCLC (without driver mutations) measurable disease by RECIST criteria, planned treatment with anti-PD-1 checkpoint inhibitor monotherapy, no clinical signs of active infection, expected survival of more than 3 months, and ability to understand and the willingness to sign a written informed consent document. Informed consent was obtained from each participating subject. The study was approved by the Institutional Ethic Committee of Masaryk Memorial Cancer Institute, reference number 2017/1890/MOU.

Assessment of clinical outcomes

The clinical treatment response was identified as stable disease (SD), progressive disease (PD), partial response (PR), or complete response (CR), according to the iRECIST. Patients were categorized into 2 groups according to disease control (DC). Patients in the DC group achieved CR, PR, or SD. Patients in the no disease control (nDC) group achieved PD as the best response to treatment. Progression-free survival (PFS) was defined as the time from the beginning of ICI therapy to the first documented objective disease progression or death from any cause. Overall survival (OS) was defined as the time from the beginning of ICI therapy to death from any cause. For the analysis of PFS, times for patients who were alive and had no disease progression or who were lost to follow-up were censored at the time of the last disease assessment. To analyze OS, data for patients who were alive or lost to follow-up were censored at the time of the last contact.

Sample collection and flow cytometry

Venous blood was collected in a 2.6 mL S-Monovette^® tube with K3EDTA anticoagulant and processed in the laboratory within the same day. A white blood cell count (WBC) and a white blood cell differential, including relative and absolute counts of lymphocytes, monocytes, neutrophils, eosinophils, and basophils, were measured by a Sysmex XN hematology analyzer. The neutrophil-to-lymphocyte ratio (NLR) was calculated as the absolute neutrophil count to the absolute lymphocyte count (ALC). Lymphocyte-to-monocyte ratio (LMR) was calculated as ALC to the absolute monocyte count. Peripheral blood circulating immune subsets were analyzed by multiparameter multicolor flow cytometer and software (Navios, Beckman Coulter). Whole blood (25 µL) was incubated with 10 µL of antibody cocktail for 15 min in the dark at room temperature, hemolysed by Versalyse^® (Beckman Coulter) for 15 min, and measured by the appropriate flow cytometric protocol. Immuno-trol Cells (Beckman Coulter) were used as daily internal quality control and daily verification of the cytometer optical alignment and fluidics system by Flow-Check Pro Fluorospheres (Beckman Coulter) was performed. Antibody cocktail compositions, antibody details, and immune cell subsets are summarised in Supplementary Table 1. Absolute cell subset counts were derived from WBC quantified by the hematology analyzer Sysmex XN series.

Statistical analysis

Patient characteristics and peripheral blood circulating immune subset levels were described using standard summary statistics, including the median and interquartile range (IQR) and mean with standard deviation (SD) for continuous variables and frequencies and proportions for categorical variables. A logistic regression model was used to evaluate the association between peripheral blood circulating immune subset levels and DC and was interpreted using the odds ratios (ORs). The adjustment for baseline levels was used to assess early changes in peripheral blood circulating immune subsets, and adjusted ORs (aORs) were used for interpretation. The optimal cut-off values to categorize patients with high and low DC rates were established using receiver operating characteristic (ROC) analysis with the criterion based on Youden’s index. For OS and PFS, survival curves were estimated using the Kaplan–Meier method and compared using the log-rank test. The Cox proportional hazard model was used to calculate hazard ratios (HRs). The follow-up was determined using the reverse Kaplan–Meier method. All statistical analyses were performed using R version 4.3.1 [9] and a common significance level 0.05.

Results

Patient characteristics of the study cohort

A total of 41 patients (median age 66 years, 10 (24%) women) were enrolled in this study. The patients were treated for MM (24 patients, 59%) or NSCLC (17 patients, 41%) in first-line (29 patients, 71%), second-line (10 patients, 24%), and third or later 2 patients (5%). During a median follow-up period of 33 months (95% CI 31–not reached), 26 (63%) patients died. The median PFS was 6.4 months (95% CI 5.1–15) with a 3-year PFS of 18% (95% CI 8.8–36%), and the median OS was 24 months (95% CI 17–not reached) with a 3-year OS of 33% (95% CI 20–53%). The best response to ICI treatment was CR in 5 patients (12%), PR in 12 (29%), SD in 8 (20%) and PD in 16 (39%). Twenty-five (61%) patients achieved DC. No differences in DC rate were observed between histological types (p = 0.680) or treatment lines (p > 0.999). All patients’ characteristics are summarised in Table 1.

Table 1.

Patient characteristics

	Overall N = 41	Malignant melanoma N = 24	NSCLC N = 17	p-value
Sex				> 0.999
Female	10 (24%)	6 (25%)	4 (24%)
Male	31 (76%)	18 (75%)	13 (76%)
Age (years)				0.596
Mean (SD)	66 (11)	67 (12)	64 (10)
Median (IQR)	67 (58, 74)	68 (59, 77)	67 (58, 72)
Range	39, 89	39, 89	42, 78
Line of treatment				< 0.001
1	29 (71%)	23 (96%)	6 (35%)
2	10 (24%)	0 (0%)	10 (59%)
> 2	2 (5%)	1 (4%)	1 (6%)
Best Response				0.523
CR	5 (12%)	4 (17%)	1 (5.9%)
PR	12 (29%)	5 (21%)	7 (41%)
SD	8 (20%)	5 (21%)	3 (18%)
PD	16 (39%)	10 (42%)	6 (35%)
DC	25 (61%)	14 (58%)	11 (65%)	0.680
Lung mts	27 (66%)	13 (54%)	14 (82%)	0.061
Liver mts	11 (27%)	9 (38%)	2 (12%)	0.085
Brain mts	2 (4.9%)	1 (4.2%)	1 (5.9%)	> 0.999
Other mts	33 (80%)	20 (83%)	13 (76%)	0.698
Radiotherapy	10 (24%)	6 (25%)	4 (24%)	> 0.999
iAE	11 (28%)	8 (35%)	3 (18%)	0.297
Missing	1	1	0
OS				0.521
Median (95% CI)	24.1 (16.9, NR)	24.8 (16.9, NR)	21.9 (11.8, NR)
3-year	33% (20%, 53%)	34% (19%, 61%)	NR
PFS				0.348
Median (95% CI)	6.4 (5.1, 14.7)	7.3 (4.8, 25.3)	6.4 (5.0, 18.6)
3-year	18% (8.8%, 36%)	23% (11%, 50%)	NR

Abbreviations CR, complete response; PR, partial response; SD, stable disease; PD, progresive disease; DC disease control; mts, metastases; irAE immune-related adverse events; NR, not reached

Association of baseline levels and early-on changes in circulating immune subsets with disease control rate

Baseline values of peripheral blood immune cell subsets by the DC are summarised in Table 2. Patients who achieved DC had a significantly lower baseline proportion of CD4 + TEM (OR 0.75; 95% CI 0.58–0.92; p = 0.013) and a borderline significantly higher baseline proportion of CD4 + TCM (OR 1.12; 95% CI 1.00–1.27; p = 0.059). (Fig. 1 A, B). Adjusted analyses for histology type and treatment line are shown in Supplementary Table 2. Additional analyses conducted separately for MM and NSCLC subgroups are included as Supplementary Tables 3 and 4.

Table 2.

Association of baseline levels in peripheral blood circulating immune subsets with disease control rate

	nDC, N = 16	DC, N = 25	OR (95% CI)	p-value
WBC (×10 9 /L)			1.07 (0.84,1.40)	0.592
Mean (SD)	7.45 (2.38)	7.89 (2.71)
Median (IQR)	6.81 (5.62, 8.57)	8.11 (5.71, 9.15)
Neutrophil count (×10 9 /L)			1.08 (0.78,1.53)	0.656
Mean (SD)	4.89 (1.91)	5.17 (2.10)
Median (IQR)	4.29 (3.83, 5.51)	4.63 (3.51, 6.19)
ALC (×10 9 /L)			0.99 (0.55,1.89)	0.964
Mean (SD)	1.78 (1.50)	1.76 (0.80)
Median (IQR)	1.25 (1.08, 2.04)	1.56 (1.33, 2.05)
NLR			0.84 (0.58,1.17)	0.320
Mean (SD)	3.89 (2.68)	3.26 (1.26)
Median (IQR)	3.23 (2.19, 4.17)	3.36 (2.45, 4.32)
LMR			0.91 (0.62,1.27)	0.548
Mean (SD)	2.99 (2.62)	2.62 (1.41)
Median (IQR)	2.49 (1.67, 3.10)	2.39 (1.73, 2.88)
MDSC count (×10 9 /L)			3.23 (0.19,67.3)	0.426
Mean (SD)	0.41 (0.19)	0.47 (0.25)
Median (IQR)	0.44 (0.28, 0.53)	0.49 (0.28, 0.63)
T-cell count (×10 9 /L)			1.79 (0.53,7.54)	0.378
Mean (SD)	1.01 (0.49)	1.16 (0.57)
Median (IQR)	0.83 (0.70, 1.33)	1.03 (0.80, 1.33)
Th-cell count (×10 9 /L)			4.61 (0.35,98.2)	0.274
Mean (SD)	0.56 (0.23)	0.65 (0.27)
Median (IQR)	0.53 (0.42, 0.60)	0.60 (0.45, 0.71)
Tc-cell count (×10 9 /L)			1.65 (0.25,15.4)	0.616
Mean (SD)	0.40 (0.28)	0.45 (0.38)
Median (IQR)	0.27 (0.21, 0.57)	0.38 (0.19, 0.57)
CD4 + TCM (% of TL)			1.12 (1.00,1.27)	0.059
Mean (SD)	15.6 (6.1)	19.5 (6.0)
Median (IQR)	16.7 (12.0, 19.9)	19.0 (14.6, 23.3)
CD4 + TEM (% of TL)			0.75 (0.58,0.92)	0.013
Mean (SD)	8.3 (4.7)	5.0 (2.2)
Median (IQR)	7.9 (4.9, 11.2)	4.6 (2.9, 6.0)
CD4 + TN (% of TL)			1.04 (0.95,1.15)	0.453
Mean (SD)	12 (7)	13 (7)
Median (IQR)	9 (7, 15)	14 (7, 19)
CD8 + TCM (% of TL)			1.14 (0.89,1.53)	0.324
Mean (SD)	4.10 (2.03)	4.96 (3.09)
Median (IQR)	3.50 (2.85, 5.34)	4.87 (2.53, 5.85)
CD8 + TEM (% of TL)			1.07 (0.74,1.60)	0.741
Mean (SD)	1.89 (1.34)	2.07 (1.95)
Median (IQR)	1.63 (0.71, 2.61)	1.57 (0.59, 2.84)
CD8 + TN (% of TL)			1.01 (0.82,1.26)	0.934
Mean (SD)	5.8 (3.4)	5.9 (2.9)
Median (IQR)	5.4 (2.6, 7.8)	5.3 (4.3, 6.9)
γδ T-cell count (×10 6 /L)			1.00 (0.99,1.01)	0.992
Mean (SD)	52 (55)	52 (46)
Median (IQR)	27 (11, 86)	36 (18, 68)
B-cell count (×10 9 /L)			0.54 (0.02,1.71)	0.414
Mean (SD)	0.34 (0.98)	0.15 (0.10)
Median (IQR)	0.08 (0.06, 0.13)	0.11 (0.08, 0.18)
Treg count (×10 6 /L)			1.02 (0.98,1.06)	0.435
Mean (SD)	32 (18)	36 (17)
Median (IQR)	24 (21, 35)	34 (21, 43)
NK-cell count (×10 9 /L)			1.11 (0.12,13.4)	0.927
Mean (SD)	0.33 (0.36)	0.34 (0.22)
Median (IQR)	0.21 (0.12, 0.35)	0.30 (0.16, 0.42)
NKT-like cell count (×10 9 /L)			7.13 (0.03,6,502)	0.521
Mean (SD)	0.11 (0.07)	0.13 (0.13)
Median (IQR)	0.10 (0.05, 0.19)	0.09 (0.05, 0.21)
CD28 + ICOS + T cells (% of TL)			1.02 (0.97,1.09)	0.486
Mean (SD)	8 (11)	11 (13)
Median (IQR)	6 (1, 9)	7 (1, 16)
Unknown	0	2
CD28 + ICOS- T cells (% of TL)			0.99 (0.94,1.05)	0.789
Mean (SD)	19 (12)	18 (13)
Median (IQR)	18 (11, 30)	18 (8, 26)
Unknown	0	2

Abbreviations OR – odds ratio, nDC – no disease control, DC – disease control, WBC, white blood count; ALC, absolute lymphocyte count; NLR, neutrophil-to-lymphocyte ratio; ALC, lymphocyte-to-monocyte ratio; TL, total lymphocytes; Th cells, helper T cells; Tc cells, cytotoxic T cells; TCM, central memory T-cells; TEM, effector memory T-cells; TN, naïve T-cells; Treg, regulatory T-cells; NK, natural killers; NKT-like, natural killers-like T-cells; MDSCs, myeloid-derived suppressor cells; ICOS, inducible T-cell costimulator

Fig. 1.

Baseline and before 2nd dose of ICI levels for A) proportion CD4 + TCM, B) proportion of CD4 + TEM, C) proportion of CD4 + TN, D) Treg count, E) proportion of CD28 + ICOS-T cells by DC. Abbreviation: OR – odds ratio, aOR – adjusted odds ratio for baseline level, nDC – no disease control, DC – disease control; TCM, central memory T-cells; TEM, effector memory T-cells; TN, naïve T-cells; Treg, regulatory T-cells; ICOS, inducible T-cell costimulator

Early-on treatment (before the second ICI dose) changes in circulating immune subsets by DC are summarised in Table 3. The increases adjusted to the baseline values in Treg count (aOR 1.10; 95% CI 1.03–1.21; p = 0.012) (Fig. 1D), proportion of CD4 + TN (aOR 1.75; 95% CI 1.24–2.92; p = 0.008) (Fig. 1C), and proportion of CD28 + ICOS- T cells (aOR 1.17; 95% CI 1.05–1.34; p = 0.012) (Fig. 1E) were significantly associated with DC. Analyses additionally adjusted for histology type and treatment line are shown in Supplementary Table 5. Separate analyses for MM and NSCLC subgroups are included as Supplementary Tables 6 and 7.

Table 3.

Association of early-on changes in peripheral blood circulating immune subsets with disease control rate

	nDC, N = 16	DC, N = 25	aOR (95% CI)	p-value
ΔWBC (×10 9 /L)			1.03 (0.58,1.82)	0.923
Mean (SD)	-0.13 (0.76)	-0.25 (1.78)
Median (IQR)	-0.36 (-0.66, 0.32)	0.02 (-1.37, 0.62)
ΔNeutrophil count (×10 9 /L)			0.98 (0.52,1.87)	0.954
Mean (SD)	-0.09 (0.64)	-0.20 (1.48)
Median (IQR)	-0.05 (-0.56, 0.45)	-0.20 (-1.00, 0.61)
ΔALC (×10 9 /L)			0.78 (0.08,7.39)	0.823
Mean (SD)	-0.01 (0.23)	-0.02 (0.37)
Median (IQR)	0.02 (-0.08, 0.04)	-0.10 (-0.19, 0.22)
ΔNLR			1.01 (0.45,2.27)	0.982
Mean (SD)	-0.13 (0.56)	-0.08 (0.96)
Median (IQR)	-0.07 (-0.28, 0.12)	0.09 (-0.57, 0.33)
ΔLMR			0.78 (0.29,2.02)	0.606
Mean (SD)	0.12 (0.53)	0.05 (0.79)
Median (IQR)	0.16 (-0.03, 0.42)	-0.13 (-0.43, 0.36)
ΔMDSC count (×10 9 /L)			7.04 (0.06,1,291)	0.424
Mean (SD)	-0.05 (0.12)	-0.04 (0.18)
Median (IQR)	-0.05 (-0.11, 0.02)	-0.03 (-0.06, 0.08)
ΔT-cell count (×10 9 /L)			5.01 (0.11,367)	0.422
Mean (SD)	0.01 (0.16)	0.02 (0.23)
Median (IQR)	-0.01 (-0.04, 0.06)	0.04 (-0.09, 0.17)
ΔTh-cell count (×10 9 /L)			73.7 (0.17,121,393)	0.201
Mean (SD)	0.00 (0.10)	0.04 (0.13)
Median (IQR)	-0.03 (-0.05, 0.04)	0.02 (-0.04, 0.12)
ΔTc-cell count (×10 9 /L)			0.04 (0.00,117)	0.440
Mean (SD)	0.01 (0.06)	-0.02 (0.12)
Median (IQR)	0.01 (-0.01, 0.03)	-0.01 (-0.09, 0.05)
ΔCD4 + TCM (% of TL)			1.03 (0.83,1.29)	0.791
Mean (SD)	0.7 (3.0)	0.5 (3.5)
Median (IQR)	0.4 (-1.0, 1.4)	1.0 (-1.3, 2.5)
ΔCD4 + TEM (% of TL)			0.68 (0.37,1.12)	0.174
Mean (SD)	0.69 (2.21)	-0.14 (1.19)
Median (IQR)	0.38 (-0.42, 1.27)	-0.07 (-0.74, 0.31)
ΔCD4 + TN (% of TL)			1.75 (1.24,2.92)	0.008
Mean (SD)	-1.9 (2.9)	1.1 (3.3)
Median (IQR)	-1.5 (-2.6, -0.9)	0.8 (0.0, 3.4)
ΔCD8 + TCM (% of TL)			0.78 (0.45,1.27)	0.337
Mean (SD)	0.47 (1.37)	0.05 (1.39)
Median (IQR)	0.03 (-0.31, 0.69)	0.16 (-0.44, 0.45)
ΔCD8 + TEM (% of TL)			0.89 (0.42,1.92)	0.754
Mean (SD)	0.21 (0.38)	0.12 (1.09)
Median (IQR)	0.13 (-0.01, 0.53)	-0.13 (-0.28, 0.13)
ΔCD8 + TN (% of TL)			1.24 (0.78,2.16)	0.393
Mean (SD)	-0.35 (1.25)	0.02 (1.63)
Median (IQR)	-0.32 (-0.78, 0.23)	0.16 (-0.55, 0.63)
Δγδ T-cell count (×10 6 /L)			1.01 (0.98,1.04)	0.656
Mean (SD)	-4 (12)	0 (33)
Median (IQR)	-2 (-6, 1)	-2 (-9, 4)
ΔB-cell count (×10 9 /L)			0.58 (0.00,2,254,543)	0.943
Mean (SD)	-0.01 (0.03)	-0.01 (0.05)
Median (IQR)	-0.01 (-0.02, 0.02)	0.00 (-0.04, 0.02)
ΔTreg count (×10 6 /L)			1.10 (1.03,1.21)	0.012
Mean (SD)	-1 (11)	8 (13)
Median (IQR)	0 (-9, 10)	7 (1, 16)
ΔNK-cell count (×10 9 /L)			0.21 (0.00,797)	0.706
Mean (SD)	-0.01 (0.09)	-0.02 (0.11)
Median (IQR)	0.00 (-0.04, 0.04)	-0.02 (-0.07, 0.04)
ΔNKT-like cell count (×10 9 /L)			0.00 (0.00,14.8)	0.150
Mean (SD)	0.02 (0.03)	-0.01 (0.06)
Median (IQR)	0.01 (0.00, 0.02)	-0.01 (-0.04, 0.01)
ΔCD28 + ICOS + T cells (% of TL)			0.94 (0.82,1.06)	0.340
Mean (SD)	-2 (13)	-6 (13)
Median (IQR)	1 (0, 3)	0 (-6, 1)
Unknown	0	2
ΔCD28 + ICOS- T cells (% of TL)			1.17 (1.05,1.34)	0.012
Mean (SD)	-2 (11)	5 (11)
Median (IQR)	-7 (-9, 1)	6 (-1, 10)
Unknown	0	2

Abbreviations aOR – adjusted odds ratio for baseline level, nDC – no disease control, DC – disease control, WBC, white blood count; ALC, absolute lymphocyte count; NLR, neutrophil-to-lymphocyte ratio; ALC, lymphocyte-to-monocyte ratio; TL, total lymphocytes; Th cells, helper T cells; Tc cells, cytotoxic T cells; TCM, central memory T-cells; TEM, effector memory T-cells; TN, naïve T-cells; Treg, regulatory T-cells; NK, natural killers; NKT-like, natural killers-like T-cells; MDSCs, myeloid-derived suppressor cells; ICOS, inducible T-cell costimulator; Δ – the absolute value of difference

Based on ROC analyses (Supplementary Fig. 1), the optimal cut-offs of 13.7% for the baseline proportion of CD4 + TCM, 8.5% for the baseline proportion of CD4 + TEM, and 13.3% for the proportion of CD4 + TN, 31 × 10⁶/L for Treg count, and 24% for the proportion of CD28 + ICOS-Tcells before the second dose of ICI were established to be of greatest utility to classify patients to groups with low and high DC rates.

Association of peripheral blood circulating immune subsets with survival outcomes

Patients with a high baseline proportion of CD4 + TCM and a low baseline proportion of CD4 + TEM showed significantly longer PFS (median of 10.3 vs. 4.6 months, p = 0.011, HR 2.6, Fig. 2A and median of 10.5 vs. 3.9 months, p ˂ 0.001, HR 0.23, Fig. 2B, respectively) and also longer OS (median of 26.3 vs. 11.7 months, p = 0.002, HR 3.75, Fig. 3A and median of 25.3 vs. 9.8 months, p ˂ 0.001, HR 0.15, Fig. 3B, respectively). Regarding the peripheral blood immune subsets before the second dose of ICI, we found longer PFS in patients with a high proportion of CD28 + ICOS- T cells (median of 21.5 vs. 5.1 months, p = 0.017, HR 2.51, Fig. 2E) and with a high Treg count (median of 13.4 vs. 4.9 months, p = 0.016, HR 2.33, Fig. 2D). Patients with a high proportion of CD28 + ICOS- T cells before the second dose of ICI also had significantly prolonged OS (median of 38.4 vs. 19.7 months, p = 0.003, HR 2.69, Fig. 3E). The high proportion of CD4 + TN before the second dose of ICI did not significantly prolong PFS (median of 11.0 vs. 5.1 months, p = 0.100, HR 1.82, Fig. 3C); likewise, the difference in OS was at the borderline of significance (31.0 vs. 19.7 months, p = 0.057, HR 2.24, Fig. 3C).

Fig. 2.

Kaplan–Meier estimates of progression-free survival (PFS) according to baseline levels for A) proportion of CD4 + TCM, B) proportion of CD4 + TEM, and before 2nd dose of ICI levels for C) proportion of CD4 + TN, D) Treg count, E) proportion of CD28 + ICOS-T cells. Abbreviations: TCM, central memory T-cells; TEM, effector memory T-cells; TN, naïve T-cells; Treg, regulatory T-cells; ICOS, inducible T-cell costimulator; mo, months

Fig. 3.

Kaplan–Meier estimates of overall survival (OS) according to baseline levels for A) proportion of CD4 + TCM, B) proportion of CD4 + TEM, and before 2nd dose of ICI levels for C) proportion of CD4 + TN, D) Treg count, E) proportion of CD28 + ICOS- T cells. Abbreviations: TCM, central memory T-cells; TEM, effector memory T-cells; TN, naïve T-cells; Treg, regulatory T-cells; ICOS, inducible T-cell costimulator; mo, months

Discussion

The ICI treatment has revolutionized the prognosis of patients with advanced MM and NSCLC. As evidenced by the CheckMate 067 study, patients with metastatic MM treated with PD-1 inhibitors reached a 5-year OS rate of 44% [1]. Similarly, promising results were shown in the 5-year follow-up of the KEYNOTE-024 trial. In this study, patients with metastatic NSCLC treated with PD-1 inhibitors achieved a 5-year OS rate of 31.9% [10]. Unfortunately, there is currently no predictive marker to navigate the decision-making process in patients with metastatic MM. Regarding the NSCLC, the most significant benefit from therapy with a PD-1 inhibitor was found in patients with high PD-L1 expression (≥ 50%) [10], and still, half of the patients in this highly selected population did not achieve a significant therapeutic response [10]. In our study, MM and NSCLC were selected for investigation as they are considered high mutation burden cancer types and typically show high initial ICI responses [11].

No relevant peripheral blood-based predictors for ICI treatment have yet been approved by the FDA or European Medicines Agency (EMA). Currently, available biomarkers are conditioned by tissue biopsy, which limits their accessibility and is compromised by tumor heterogeneity. Therefore, peripheral blood-based circulating immune subsets may be a promising non-invasive biomarker for patients treated with ICIs. Moreover, inflammatory indexes such as NLR and LMR were investigated, as they can reflect antitumor immunity and, thus, serve as potential predictive markers for ICI therapy. While numerous studies have indicated that a higher NLR and lower LMR are associated with poorer outcomes and reduced response to ICIs [12–14], some research, as well as ours, has shown no significant correlation between NLR [15, 16], LMR [17–19] and response to immunotherapy. These contradictory results may be explained by NLR and LMR being rather prognostic than predictive factors.

For a long time, CD8 + T cells have been key players in antitumor immunity thanks to their potent cytotoxicity, enabling direct tumor-cell killing [20]. However, tumor cells can escape CD8 + T cell recognition through immunological editing [21]. On the other hand, the importance of CD4 + T cells and their subtypes for antitumor responses is less recognized and not straightforward in the context of cancer progression. Nevertheless, an increasing number of preclinical studies support the crucial role of CD4 + T cells in innate and adaptive antitumor immune responses [22] and the stronger selective pressure of mutations in major histocompatibility complex-II (MHC-II) restricted neoantigens compared to MHC-I reinforces the key contribution of CD4 + T cells in cancer immunosurveillance [23].

In our study, we observed that a pretreatment high proportion of CD4 + TCM and a low proportion of CD4 + TEM predicts response to PD-1 inhibitors and prolonged PFS and OS. These two memory T cell populations differed in the range of produced cytokines, their ability to traffic to lymphoid organs and tissues, and their efficacy in antitumor immunity. Upon a secondary challenge, TEMs represent an available pool of antigen-primed cells that mediate faster response and primarily secrete effector cytokines, such as interferon-gamma (IFNγ) or interleukin-4 (IL-4). On the other hand, TCMs lacked immediate effector capacity. Nonetheless, upon restimulation, they can still efficiently stimulate dendritic cells, help B cells, generate a new wave of effector cells, and mainly produce IL-2 [24]. Similar findings, such as low CD4 + TEM proportion or high TCM: TEM ratio associated with longer PFS, were observed in previous studies with NSCLC patients treated with immunotherapy [25, 26]. The different association of memory CD4 + cell subtypes with ICI disease control could be explained by the fact that whereas TCMs have the potential to self-renew, TEMs are more likely to have exhausted their proliferation potential [27]. Thanks to TCMs’ self-renewal potential and the ability to generate more differentiated TEMs and effector T cells [28], the high proportion of TCMs reflects the capacity of an immune system to escalate an antitumor immune response that was potentiated by ICIs and thus in line with our results predict the response to PD-1 inhibitors.

The second part of our study was focused on early-on changes in peripheral blood circulating immune subsets associated with response to PD-1 inhibitors. Our results demonstrated that an increase in Treg count predicted disease control and a high Treg count before the second dose of PD-1 inhibitors was associated with a prolonged PFS. This early-on treatment increase in Treg count during ICI therapy, similar to our findings, was also demonstrated by Kim et al. in patients with advanced NSCLC [29] and by Araujo et al. across multiple tumor types [30]. Although Tregs are considered immune suppressor cells due to their ability to negatively modulate various immune functions from initial T-cell activation to effector function in the target tissue, the association between high levels of these suppressor cells and response to immunotherapy could be explained by Treg cells expansion strictly depending upon IL-2 levels produced by specific antigen-activated effector T cells [31]. Thus, an increase in Treg could result from maintaining homeostasis upon the successful activation of T cells during ICI therapy. Indeed, in Kim et al.‘s study, cytokines involved in Tregs development and maturation, such as granulocyte macrophage-colony stimulating factor (GM-CSF), IL-12, IL-2, and IL-15, are increased in patients who have benefited from anti-PD-1 therapy [29], suggesting that Treg cells might be an indirect indicator of an antitumor immune response.

As mentioned above, CD4 + T cells are required for efficacious CD8 + responses under PD-1 inhibitors, which was proved by Zuazo et al. In their study, all systemic CD8 + T cells in patients before the start of ICI were dysfunctional, and the proliferative capacities of CD8 + T cells were recovered only in patients with functional CD4 + immunity [32]. The initial step of differentiation of the naïve cells is antigenic stimulation that eventually leads to naïve cell proliferation and differentiation into specific effector cells [33]. In our study, patients who manifested an early-on treatment elevation in the CD4 + TN fraction achieved DC. The predictive value of the high pretreatment percentage of CD4 + TN in NSCLC has been demonstrated in two previous studies [25, 34]. PD-1 blockade has been shown to affect naïve T cells [35]; as in HIV-positive patients, PD-1 blockade inhibited the transition of naïve CD4 + and CD8 + T cells to effector memory phenotypes [36]. Still, to our knowledge, our study was the first to prove that ICI-induced expansion of CD4 + TN is associated with therapeutic response to ICIs.

Although many previous ICI studies have assessed the predictive value of the expression of inhibitory receptors on CD8 + or CD4 + T cells, the expression of co-receptors (receptors for costimulatory molecules) has not been a primary focus. Over the last years, evidence has been increasing that CD28 expression is essential for PD-1 inhibitors. Our study demonstrated that therapy with PD-1 inhibitors led to an expansion of CD28 + ICOS- T cells in patients who achieved DC, and also, the early-on treatment high proportion of CD28 + ICOS- T cells was associated with longer PFS and OS. We have not found any changes in the proportion of CD28 + ICOS + T cells associated with treatment response. Importantly, as ICOS is expressed only on activated T cells [37], CD28 + ICOS- T cells overlap with the naïve T cells phenotype, and CD28 + ICOS + T cells overlap with the memory T cells phenotype. Thus, the early-on increase in CD28 + ICOS- T cells associated with response to ICI is consistent with the already mentioned early-on treatment increase of naïve T cells. We did not find any early-on treatment changes in the memory T cell subsets; likewise, there were no changes in CD28 + ICOS + T cells associated with response to ICI. Although some prior studies confirmed that the expansion of ICOS + T cells can predict the response to ICI, this has been proven only with CTLA-4 inhibitors [38, 39] but not PD-1 inhibitors [40]. Recent preclinical studies showed that the CD28 costimulatory receptor is a primary target for PD-1 inhibitors [41], and it is necessary for the reinvigoration of antitumor CD8 + T cell response to PD-1 inhibition [8]. Kamphorst et al. also confirmed these preclinical findings in NSCLC patients treated with PD-1 inhibitors, where activated CD8 + T cells by PD-1 therapy were mainly CD28+ [8].

We are aware of certain limitations of our study. Despite the prospective design and uniform treatment with PD-1 inhibitors, the study was limited by the relatively low number of enrolled patients. Furthermore, our study cohort included two distinct histological tumor types. In accordance with the treatment guidelines at the time of initiation of our study, the majority of MM patients received ICI as the first line, while NSCLC patients typically received ICI as the second line. Although the sample sizes of both histology subgroups were limited for stratified analyses, additional analyses for each histological type were provided to ensure completeness. We found no significant associations between baselines and early-on changes with DC in the NSCLC subgroup, although the pretreatment proportion of CD4 + TEM cells was borderline. In the MM subgroup, we found that patients who achieved DC showed an increased proportion of CD4 + TN and CD28 + ICOS- T cells after one dose of ICIs and tended to have a lower pretreatment proportion of CD4 + TEM cells. However, no significant changes were observed in the Treg count. The results in the histology subgroups were generally consistent with those observed in the entire cohort, further supported by the analysis adjusted for histological type and treatment line.

Conclusion

In conclusion, this study demonstrated that a low proportion of circulating CD4 + TEM and a high proportion of circulating CD4 + TCM at baseline were associated with response to PD-1 inhibitors in patients with advanced MM and NSCLC. More importantly, we have found that the early-on changes after initiation of therapy with PD-1 inhibitors, such as an increase in Treg count and the proportion of CD4 + TN and CD28 + ICOS- T cells, are associated with disease control. The high Treg count and proportion of CD28 + ICOS- T cells also showed predictive value for therapy with PD-1 inhibitors. Despite the evidence presented in this study indicating the predictive potential of blood-based immune biomarkers and broadening the understanding of immune cell dynamics induced by PD-1 inhibition, further validation in other cancer types and larger cohorts is necessary.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

ALC

Absolute Lymphocyte Count

CR

Complete Response

CRC

Colorectal Cancer

CTLA-4

Cytotoxic T lymphocyte-associated protein-4

DC

Disease Control

EMA

European Medicines Agency

FDA

Food and Drug Administration

GM-CSF

Granulocyte Macrophage-Colony Stimulating Factor

ICI

Immune Checkpoint Inhibitor

ICOS

Inducible T-cell Costimulator

IFNγ

Interferon Gamma

IL

Interleukin

IQR

Interquartile Range

irAE

immune-related Adverse Event

LMR

Lymphocyte-to-Monocyte Ratio

MDSC

Myeloid-Derived Suppressor Cell

MM

Malignant Melanoma

mo

months

MSI/dMMR

Microsatellite Instability/defective Mismatch Repair

mts

metastases

nDC

no Disease Control

NK

Natural Killers

NKT-like

Natural Killers-like T-Cell

NLR

Neutrophil-to-Lymphocyte Ratio

NR

Not Reached

NSCLC

Non-Small Cell Lung Cancer

ORs

Odds Ratios

OS

Overall Survival

PD

Progressive Disease

PD-1

Programmed Death 1

PD-L1

Programmed Death-Ligand 1

PFS

Progression-Free Survival

PR

Partial Response

ROC

Receiver Operating Characteristic

SD

Stable Disease

Tc cells

Cytotoxic T cells

TCM

Central Memory T Cell

TCR

T-Cell Receptor

TEM

Effector Memory T Cell

Th cells

Helper T cells

TL

Total Lymphocytes

TMB

Tumor Mutational Burden

TN

Naïve/Early T Cell

Treg

Regulatory T-cell

WBC

White Blood Cell Count

Author contributions

Conceptualization was done by SB, PG, IS, LG and LZD; Methodology was done by PG, LZD and LG; Formal analysis was done by IS, SB and PG; Investigation was done by SB, PG, IS, LG, BV, JN, MM, OB, RL, AP, JP, IK, RV, LZD; Writing - original draft was done by SB, PG, IS, LG, LZD and figures and tables were prepared by IS; Funding acquisition was done by PG and RV; Supervision was done by PG and LZD. All authors reviewed the results and approved the final version of the manuscript.

Funding

This work was financially supported by the Ministry of Health of the Czech Republic, grant no. NU21-03-00539, conceptual Development of Research Organization MMCI 00209805, by National Institute for Cancer Research (Programme EXCELES, ID Project No. LX22NPO5102)—Funded by the European Union—Next Generation EU and by Specific University Research (MUNI/A/1580/2023) provided by MEYS.

Data availability

The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethical approval

The study was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board and Ethics Committee of Masaryk Memorial Cancer Institute, reference number 2017/1890/MOU; 27 June 2017.

Consent to participate

Informed consent was obtained from all individual participants included in the study.

Consent for publication

Not applicable.

Competing interests

Dr. Borilova reported receiving personal fees from Bristol Myers Squibb and MSD outside the submitted work, Dr. Grell reported receiving personal fees from Bristol Myers Squibb, Roche and Servier outside the submitted work, Dr. Poprach reported receiving personal fees from Roche, Bristol Myers Squibb, Merck KGaA, MSD, Novartis, Astellas, Janssen, and Sanofi/Aventis, Ipsen, and Pfizer outside the submitted work. The other authors have no competing interests to declare that are relevant to the content of this article.