Nivolumab receptor occupancy on effector regulatory T cells predicts clinical benefit

Abstract

Immune checkpoint inhibitor discovery represents a turning point in cancer treatment. However, the response rates of solid tumors remain ~10%–30%; consequently, prognostic and immune‐related adverse event (irAE) predictors are being explored. The programmed cell death protein 1 (PD‐1) receptor occupancy (RO) of PD‐1 inhibitors depends on the number of peripheral blood lymphocytes and their PD‐1 expression levels, suggesting that the RO may be related to efficacy and adverse events. As PD‐1 inhibition affects each T‐cell subset differently, the RO of each cell population must be characterized. However, relevant data have not been reported, and the prognostic relevance of this parameter is not known. In this study, we aimed to clarify the association between the nivolumab RO in each T‐cell population and patient prognosis and reveal the development of irAEs in nivolumab‐treated patients. Thirty‐two patients were included in the study, and the mean follow‐up period was 364 days. The nivolumab RO on effector regulatory T cells (eTregs) was significantly lower in the group that presented clinical benefits, and a significant negative association was observed between PD‐1 occupancy on eTregs and all‐cause mortality. The results suggest that the nivolumab RO on eTregs may be a prognostic factor in PD‐1 inhibitor therapy, implying that the inhibition of PD‐1/PD‐ligand 1 (PD‐L1) signaling on eTregs may attenuate antitumor effects.

Measurement of PD‐1 receptor occupancy. PBMCs were saturated with either unlabeled human IgG4 (isotype control) or nivolumab and then co‐stained with murine anti‐human IgG4 biotin plus streptavidin‐phycoerythrin.

Abbreviations

12 M DCM

DCB lasting 6 months

6 M DCB

DCB lasting 6 months

APC

allophycocyanin

BV

brilliant violet

CCR7

C–C motif chemokine receptor 7

CI

confidence interval

CR

complete response

CT

computed tomography

CTCAE

Common Terminology Criteria for Adverse Events

Cy

cyanine

DCB

durable clinical benefit

eTregs

effector regulatory T cells

FoxP3

Forkhead box P3

HR

hazard ratio

ICI

immune checkpoint inhibitor

IQR

interquartile range

irAE

immune‐related adverse event

MSI‐H

microsatellite instability‐high

NCB

no clinical benefit

NLRs

neutrophil/lymphocyte ratios

OS

overall survival

PD‐1

programmed cell death protein 1

PD‐L1

programmed cell death ligand 1

PE

phycoerythrin

PFS

progression‐free survival

PR

partial response

RO

receptor occupancy

ROC

receiver operating characteristic

TILs

tumor‐infiltrating lymphocytes

TMB

tumor mutational burden

1. INTRODUCTION

Immune checkpoint inhibitor discovery has resulted in a paradigm shift in cancer treatment. However, the response rates of solid tumors remain ~10%–30%, ¹ , ² and serious irAEs have been reported. Therefore, prognostic and irAE predictors are explored. ³ , ⁴ Programmed cell death ligand 1 expression, MSI‐H, TMB, TILs, gut microbiota, and NLRs have been reported as prognostic predictors ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ ; however, their predictive capacity is limited.

The PD‐1/PD‐L1 signaling pathway is a negative signaling pathway in which the binding of PD‐1 expressed on T cells to PD‐L1 expressed on tumor cells and tumor‐associated macrophages inhibits T‐cell activation and causes cancer cells to escape tumor immunity. PD‐1 inhibitors inhibit the binding of PDL‐1 to PD‐1 and restore tumor immune responses by reducing inhibitory signals to T cells. However, PD‐1 inhibitors are considered to exert their antitumor effects primarily through CD8 T cells. ¹⁴ The inhibition of PD‐1 signaling in regulatory T cells (Tregs) causes the suppression of antitumor activity associated with Treg proliferation and activation. ¹⁵ Additionally, the antitumor active effects of PD‐1 inhibition differ by T‐cell fraction.

The doses of PD‐1 inhibitors are defined as unchanged and the same dose is administered to each patient, whereas the doses of cytotoxic anticancer agents are defined by dose‐dependent tolerability. While reports have indicated that serum nivolumab concentrations do not affect the timing and severity of irAEs, peripheral blood lymphocyte counts may affect the serum nivolumab concentrations. ¹⁶ Therefore, the PD‐1 RO of PD‐1 inhibitors depends on the number of peripheral blood lymphocytes and their PD‐1 expression, suggesting that RO may be related to efficacy and adverse events. Although several clinical trials have analyzed the RO of these drugs, the association between RO and efficacy is not clear. ¹⁷ , ¹⁸

Importantly, PD‐1 inhibitors act differently on each T‐cell subset, as each subset is individually activated and proliferates by mitosis according to the state of the tumor microenvironment. Therefore, the drug RO in each cell population must be analyzed; however, the RO of these drugs has not been previously reported and their prognostic relevance is not known. In this study, we aimed to clarify the association between the nivolumab RO in each T‐cell population and the prognosis and development of irAEs in nivolumab‐treated patients.

2. MATERIALS AND METHODS

2.1. Patients and sample collection

Participants who were diagnosed with solid cancers, including cancers of unknown primary origin, and treated with nivolumab (2 mg/kg every 2 weeks) from January 2014 to October 2017 at our institute were included. Patients who underwent treatment before the collection of peripheral blood, who did not receive a second course of nivolumab as scheduled, and who presented with autoimmune diseases requiring immunosuppressive therapy were excluded. The present study was performed per the tenets of the Declaration of Helsinki and was approved by the Ethics Committee of Showa University School of Medicine (Approval #2165). Written informed consent was obtained from all enrolled patients.

2.2. Clinical benefits and irAEs

Tumor response was assessed using CT or MRI every 12 weeks according to RECIST (version 1.1). Durable clinical benefit was defined as a CR, PR, or stable disease. The patients were grouped according to the clinical benefits as follows: NCB, 6 M DCB, and 12 M DCB. Progression‐free survival was defined as the time from the start of anti‐PD‐1 therapy to either disease progression (according to RECIST v1.1) or death from any cause. Overall survival was defined as the time from the start of anti‐PD‐1 therapy to death from any cause. IrAEs were graded per CTCAE version 4.0.

2.3. Immunological analysis

Peripheral blood was collected before treatment (day 0) and 14 days after the administration of the first dose of nivolumab. PBMCs from each patient were harvested using density gradient centrifugation with Ficoll‐Paque (Amersham Pharmacia Biotech). PBMCs were stained using Fixable Viability Dye (FVS), PerCP‐Cy‐5.5‐conjugated human CD3 Ab, PE‐Cy7‐conjugated human CD4 Ab, BV‐510‐conjugated human CD8 Ab, APC‐conjugated human FoxP3 Ab, BV650‐conjugated human CCR7 antibody (BD Biosciences, San Jose, CA, USA), and FITC‐conjugated human CD45RA antibody (Miltenyi Biotec), and were sorted into live cell (FVS⁻), followed by naive T cell T_N (CCR7⁺CD45RA⁺), central memory T cell T_CM (CCR7⁺CD45R⁻), effector memory T cell T_EM (CCR7⁻CD45RA⁻), CD45RA effector memory T cell T_EMRA (CCR7⁻CD45RA⁺), and eTreg (CD4⁺CD45RA⁻Foxp3^hi) populations (Figure S1). Cells were stained according to the manufacturer's flow cytometry preparation protocol for each antibody, and the stained cells were detected using a BD LSRFORTESSA X‐2 flow cytometer (BD Biosciences). Flow cytometric data were analyzed using FlowJo software, version 10.3.0 (FlowJo).

2.4. Measurement of PD‐1 RO

Measurement of PD‐1 RO was performed as described previously. ¹⁷ , ¹⁹ PBMCs were preincubated for 30 min at 4°C with a saturating concentration (20 μg/mL) of either unlabeled human IgG4 (isotype control) or nivolumab, washed extensively, and then co‐stained with PE‐conjugated murine anti‐human IgG4 (BD Biosciences). PD‐1 RO by the infused nivolumab was estimated as the ratio of the percent of cells stained with anti‐human IgG4 after in vitro saturation with the isotype control antibody (indicating in vivo binding) to that observed after nivolumab saturation (indicating total available binding sites) (Figure 1).

FIGURE 1.

Measurement of PD‐1 RO. PBMCs were saturated with either unlabeled human IgG4 (isotype control) or nivolumab and then co‐stained with murine anti‐human IgG4 biotin plus streptavidin‐phycoerythrin. The PD‐1 RO was estimated as the ratio of the percent of cells stained with anti‐human IgG4 after in vitro saturation with isotype control antibody (indicating percentage of in vivo nivolumab‐conjugated PD‐1) to that observed after nivolumab saturation (indicating total percentage of PD‐1). PD‐1, programmed cell death protein 1; RO, receptor occupancy.

2.5. Statistical analysis

Categorical data are described as numbers with proportion (%), and they were compared using Fisher's exact test. Continuous data are expressed as mean with SD or as median with IQR, as appropriate. During the main analysis, Cox proportional hazard regression analysis was conducted to assess the association between the RO of PD‐1 and OS after adjusting for total lymphocyte count. OS was analyzed using the Kaplan–Meier method. All analyses were performed using JMP software, version 14.0 (SAS Institute). Correlations or associations were considered to be statistically significant when the two‐tailed p‐value was less than 0.05. Three sensitivity analyses were performed for the main analysis. First, we analyzed after adjusting for NLR instead of the total lymphocyte count. Second, we analyzed only for adenocarcinomas. Third, we analyzed only lung and stomach primary lesions including effective cases at 6 months.

3. RESULTS

3.1. Patients

Patient characteristics are presented in Table 1. Thirty‐two patients were included in the study. The mean follow‐up period was 364 days, and the median PFS durations were 354 and 747 days in the 6 M and 12 M DCB groups, respectively. Most participants were men (75%), and the mean age was 68.4 years. The 6 M DCB group consisted of six patients with lung and gastric primary lesions.

TABLE 1.

Patient characteristics.

	Total (n = 32)	Clinical benefit at 6 months		Clinical benefit at 12 months
		DCB (n = 6)	NCB (n = 26)	DCB (n = 3)	NCB (n = 29)
Male (%)	24 (75)	5 (83)	19 (73)	3 (100)	21 (72)
Age, median (range)	68.5 (49–84)	62 (64–84)	67.5 (49–82)	71 (66–78)	68.5 (49–84)
Performance status, mean	0.97	0.83	1.0	0.67	1.0
The number of previous lines, mean	1.8	1.3	1.9	1.3	1.8
Total lymphocyte count, mean	1317	1336	1235	1308	1393
NLR, mean	3.34	2.63	3.51	2.43	3.44
Primary lesion
Lung	9	4	5	2	7
Gastric	17	2	15	1	16
Esophagus	4	0	4	0	4
Head and neck	1	0	1	0	1
Unknown	1	0	1	0	1
The number of metastatic sites, mean	0.84	0.67	0.88	0.67	0.86
Histologic type
Adenocarcinoma	26	5	21	2	24
Squamous cell carcinoma	6	1	5	1	5
PFS, median	70	354	67	747	68
OS, median	167	820	134	908	143
irAE ≧ Grade 3	6	4	2	2	4

Abbreviations: DCB, durable clinical benefit; irAE, immune‐related adverse events; NCB, no clinical benefit; OS, overall survival; PFS, progression‐free survival.

3.2. Clinical course of patients with solid cancers treated with nivolumab and kinetic changes in T‐cell subsets

The changes in T‐cell subsets in terms of clinical benefit were analyzed by comparing the T‐cell subsets in the NCB and DCB groups before and 14 days after the administration of the first dose of anti‐PD‐1 antibody treatment (Figure 2A,B, respectively). Before the anti‐PD‐1 antibody treatment, the proportion of the CD4⁺ T_CM population in the DCB group was significantly higher than that in the NCB group at 6 months of treatment (p = 0.0228, Figure 2A). In addition, we compared the distribution of T‐cell subsets before and after anti‐PD‐1 antibody treatment in the context of the development of irAEs (Figure 2C,D, respectively). Before anti‐PD‐1 antibody treatment, the proportion of CD4⁺ TCM was significantly higher in patients with irAEs of grade ≥3 than in patients with irAEs grade <3 (p = 0.0156; Figure 2C).

FIGURE 2.

Relationship between T‐cell subsets and clinical benefit. (A) Association between T‐cell subsets before anti‐PD‐1 antibody treatment and clinical benefit after 6 months of treatment. Left, CD8 T cells; middle, CD4 T cells; right, Treg fr2. (B) Association between T‐cell subsets after 14 days of anti‐PD‐1 antibody treatment and clinical benefit after 6 months of treatment. (C) Association between T‐cell subsets before anti‐PD‐1 antibody treatment and the development of irAEs. (D) Association between T‐cell subsets after 14 days of treatment and the development of irAEs. Statistical analyses were performed using Student's t‐test. *p < 0.05. Error bars represent mean ± SEM. DCB, durable clinical benefit; NCB, no clinical benefit; PD‐1, programmed cell death protein 1; Treg, regulatory T cell, irAEs, immune‐related adverse events. CM, central memory; EM, effector memory; TEMRA, CD45RA effector memory.

3.3. Clinical course of patients with solid cancers treated with nivolumab and PD‐1 expression by T‐cell subset

The PD‐1 positivity of each T‐cell subset before and 14 days after the administration of the first dose of anti‐PD‐1 antibody treatment was examined in the context of the clinical benefit of the treatment (Figure 3A,B, respectively). We also compared the PD‐1 positivity in each T‐cell subset before and at 14 days after the anti‐PD‐1 antibody treatment in the context of irAE development (Figure 3C,D, respectively). After 14 days of treatment, the PD‐1 positivity in the CD4⁺ T_CM and CD4⁺ T_EMRA populations was significantly lower in the DCB group than in the NCB group (p = 0.0205, p = 0.0025, respectively; Figure 3B). The PD‐1 positivity in each T‐cell subset was not significantly associated with the development of irAEs.

FIGURE 3.

Relationship between PD‐1 expression on memory T cells and clinical benefit. (A) Association between PD‐1 expression on T‐cell subsets before anti‐PD‐1 antibody treatment and clinical benefit after 6 months of treatment. Left: CD8 T cells; middle: CD4 T cells; right: eTregs. (B) Association between PD‐1 expression on T‐cell subsets after 14 days of anti‐PD‐1 antibody treatment and clinical benefit after 6 months of treatment. (C) Association between PD‐1 expression on T‐cell subsets before anti‐PD‐1 antibody treatment and the development of irAE. (D) Association between PD‐1 expression on T‐cell subsets after 14 days of treatment and the development of irAEs. Statistical analyses were performed using Student's t‐test. *p < 0.05. Error bars represent mean ± SEM. DCB, durable clinical benefit; eTregs, effector regulatory T cells, irAEs, immune‐related adverse events; NCB, no clinical benefit; PD‐1, programmed cell death protein 1. CM, central memory; EM, effector memory; TEMRA, CD45RA effector memory.

3.4. Clinical course of patients with solid cancers treated with nivolumab and PD‐1 RO on T‐cell subsets

After adjusting for the total lymphocyte count, the Cox proportional hazard regression analysis showed a significant association between PD‐1 occupancy on eTregs and all‐cause mortality HR 1.421; 95% CI 1.072–1.943; Table 2). In the first sensitivity analysis, after adjusting for NLR instead of the total lymphocyte count, the Cox proportional hazard regression analysis showed a significant association between PD‐1 occupancy on eTregs and all‐cause mortality (HR 1. 411; 95% CI 1.026–2.007; p = 0. 0337). According to the second sensitivity analysis, we analyzed only for adenocarcinomas and after adjusting for the total lymphocyte count, the Cox proportional hazard regression analysis showed a significant association between PD‐1 occupancy on eTregs and all‐cause mortality (HR 1.400; 95% CI 1.039–1.975; p = 0.0259). During the third sensitivity analysis, we analyzed only lung and stomach primary lesions including effective cases at 6 months and after adjusting for the total lymphocyte count, the Cox proportional hazard regression analysis showed a significant association between PD‐1 occupancy on eTregs and all‐cause mortality (HR 1.406; 95% CI 1.016–2.030; p = 0.0395). Furthermore, we analyzed only lung and gastric primary lesions after adjusting for the primary lesion (lung or gastric). The Cox proportional hazard regression analysis showed a significant association between PD‐1 occupancy on eTregs and all‐cause mortality (HR 1. 373; 95% CI 1.040–1.884; p = 0.0242). Results of univariate analysis of associations for PD‐1 occupancy on and all‐cause mortality are presented in Table S1. No significant association was found between eTreg RO and clinical background (Figure S2). The PD‐1 occupancy on eTregs was significantly lower in the DCB group than in the NCB group at both 6 and 12 months of treatment (p = 0.0105, p = 0.0486, respectively; Figure 4A). A trend toward lower PD‐1 occupancy on eTregs was observed in patients with irAEs of grade ≥3, but this difference was not significant (Figure 4B). Using the ROC curve of PD‐1 occupancy on eTregs with clinical benefit at 6 months as the outcome, the cutoff was found to be 76.9% (Figure 4C). Low PD‐1 occupancy (<76.9%) on eTregs implied a significant prolongation in OS compared with high PD‐1 occupancy (≥76.9%) (HR 5.02, 95% CI 1.76–14.32; Figure 4D). There was no significant difference in PS, the number of previous lines, and the number of metastatic sites between the two groups of high and low PD‐1 occupancy on eTregs (Figure S3). The same analysis as for eTreg in CD8⁺ T_CM showed no significant differences (Figure S4). These results indicate that PD‐1 occupancy on eTregs is a prognostic factor for PD‐1 antibody therapy.

TABLE 2.

Overall survival of patients with solid cancers treated with nivolumab and PD‐1 RO on T‐cell subsets.

	PD‐1 RO (%)	HR	95% CI	p‐value
CD8 TN	85.1	1.220	0.935–1.660	0.5237
CD8 TCM	86.3	1.414	0.999–2.998	0.0616
CD8 TEM	89.1	1.695	0.939–3.263	0.0960
CD8 TEMRA	84.7	1.227	0.841–1.822	0.2906
CD4 TN	86.2	1.122	0.881–1.471	0.3744
CD4 TCM	82.2	1.120	0.764–1.673	0.5677
CD4 TEM	85.8	1.386	0.777–2.476	0.2884
CD4 TEMRA	88.5	1.226	0.842–1.861	0.3091
eTreg	77.4	1.421	1.072–1.943	0.0196

Abbreviations: CI, confidence interval; eTreg, effector regulatory T cell; HR, hazard ratio for 10% increase; PD‐1, programmed cell death protein 1; RO, Receptor occupancy; T_N, naive T cell; T_CM, T central memory T cell; T_EM, effector memory T cell; T_EMRA, CD45RA effector memory T cell.

FIGURE 4.

Relationship between PD‐1 RO on eTregs and clinical benefit and occurrence of immune‐related adverse events. (A) Association between the RO of PD‐1 on eTregs and clinical benefit. left: at 6 months of treatment; right: at 12 months of treatment. (B) Association between the RO of PD‐1 on eTregs and immune‐related adverse events. (C) ROC curve of PD‐1 occupancy on eTregs with clinical benefit at 6 months of treatment as the outcome. (D) OS in cases with low PD‐1 occupancy (<76.9%) and high PD‐1 occupancy (≥76.9%) on eTregs. Statistical analyses were performed using Student's t‐test. OS was analyzed using the Kaplan–Meier method. AUC, area under the curve; CI, confidence interval; DCB, durable clinical benefit; eTregs, effector regulatory T cells, ROC, receiver operating characteristic; HR, hazard ratio; n.s., not significant; NCB, no clinical benefit; OS, overall survival; PD‐1, programmed cell death protein 1; RO, receptor occupancy.

4. DISCUSSION

Here, nivolumab RO on eTregs was significantly lower in the group that achieved clinical benefit than in the group that did not achieve clinical benefit, and a significant association was found between PD‐1 occupancy on eTregs and all‐cause mortality. These findings suggest that RO on eTregs may be a prognostic factor in PD‐1 inhibitor therapy. Tregs are a type of CD4⁺ T cells that negatively regulate the immune system to prevent autoimmune diseases and accidental destruction of healthy cells. They work to maintain immune system homeostasis. They are important for mounting immune responses based on the transendocytosis uptake of CD80 on antigen‐presenting cells through CTLA‐4 and IL‐10 expression; this suppresses adverse immune responses such as autoimmunity, allergy, and excessive inflammation, and is important for immune tolerance and immune homeostasis. However, these actions may interfere with antitumor immunity. ²⁰ , ²¹ , ²² , ²³ A study on tumor‐infiltrating T cells in patients with lung and gastric cancers treated with PD‐1 inhibitors reported that PD‐1 inhibitor treatment was successful in cases with high PD‐1 expression on effector T cells and low PD‐1 expression on Tregs. ²⁴ Hyperprogressive disease is considered a rapidly exacerbating condition that can occur after PD‐1 inhibitor treatment, and an association between TILs and PBMC Tregs and the development of hyperprogressive disease has also been reported. ¹⁵ , ²⁵ In other words, treatment with anti‐PD‐1 antibodies suppresses the antitumor activity of CD8⁺ T cells by activating Tregs. Therefore, we measured PD‐1 RO on T‐cell fractions in PBMCs, because tumor‐infiltrating T cells cannot be used as a biomarker in clinical practice. Interestingly, although all T‐cell fractions in the peripheral blood should have the same exposure to the antibody at the time of initial nivolumab administration, 14 days after administration, not all T‐cell fractions were equally occupied with nivolumab, and PD‐1 RO differed for each subset of T cells. Furthermore, its PD‐1 RO differed from case to case. This fact may explain why the therapeutic effect of anti‐PD‐1 antibody is not dose dependent. ²⁶ The observations could be explained by the dilution of RO due to cell proliferation and division, as T‐cell subsets are individually differentiated and proliferate in various phases. It has been reported that CD25⁺CD4⁻ regulatory T cells have a slower division rate than CD25⁻CD4⁺ T cells under the same conditions in vitro. ²⁷ It was suggested that the dilution of RO occurred because effector T cells exert their action with cell division. Conversely, Tregs were less divided than effector T cells, eTregs may remained receptor‐occupied and their function was activated. These hypotheses need to be further investigated. Furthermore, reports have indicated that serum nivolumab levels vary with peripheral blood lymphocyte counts ¹⁶ and that these lymphocyte counts are associated with prognosis. ¹³ We performed a multivariate analysis adjusting for the peripheral blood lymphocyte count as a confounding factor. The results showed that a higher RO of PD‐1 on eTregs was associated with worse OS and lower clinical benefit. In our future studies, we will measure the changes in PD‐1 antibody binding over time in patients who had discontinued PD‐1 antibody treatment due to irAEs.

The RO of PD‐1 on CD8 and CD4 T cells, categorized into four subsets by T‐cell differentiation markers CD45RA, and CCR7, was not effective in predicting prognosis. The memory T‐cell fraction among PBMCs has been reported to fluctuate before and after PD‐1 antibody treatment, but whether this is prognostic or predictive varies among studies is not known. ²⁸ , ²⁹ , ³⁰ A positive correlation has been reported between the T_CM fraction of transplanted cells and the prognosis in adoptive immunotherapy. The proliferative potential of immature T cells is important in immunotherapy and is expected to have long‐lasting effects. ³¹ , ³² The activation of memory T cells due to the reduction in PD‐1 inhibition may recapitulate the activated peripheral T‐cell state in both adoptive immune cell types. In our study, the pretreatment proportions of CD4⁺ T_CM were significantly higher and the proportions of CD4⁺ T_EM and CD4⁺ T_EMRA tended to be lower in the group that showed a response, but this difference was not significant, which is consistent with the importance of immature T cells in the efficacy of adoptive immuno‐cell therapy. The CD8⁺ T_EM and CD4⁺ T_EM fractions before PD‐1 inhibition therapy tended to be lower in the responding group than in the non‐responding group; this trend was reversed after the treatment, and the CD8⁺ T_EM and CD4⁺ T_EM fractions tended to be higher in the responding group than in the non‐responding group. This finding is interesting because it suggests that PD‐1 inhibition in the responder group promoted the differentiation of the T_CM population into T_EM by relieving T_CM exhaustion. After treatment, the response group tended to have lower PD‐1 expression in all memory T‐cell fractions than the non‐responsive group, which indicates amelioration of this exhaustion. The memory fraction tended to exhibit a positive correlation between PD‐1 RO and OS in CD8⁺ T_CM and CD8⁺ T_EM, although no significant difference was observed. We plan to increase the number of cases in the future to enable the observation of robust facts.

PD‐1 is a type I transmembrane receptor expressed on activated T cells, B cells, and myeloid cells, but mostly on T cells. It is not known whether there are differences in antibody affinity between myeloid and Treg cells. However, T cells with low affinity for tumor antigens preferentially proliferated in the PD‐1 signaling pathway. In other words, it has been reported that PD‐1 qualitatively regulates T‐cell responses by preferentially suppressing T cells with low affinity. ³³ Therefore, when T cells compete with anti‐PD‐1 antibodies, PD‐1 on T cells may also depend on the expression level of PD‐1 in response to the antibody. On the other hand, the above difference in affinity may be due to differences in signaling pathways from each TCR, which requires further investigation.

This study had two limitations. First, the RO in tumor tissue was not examined. Peripheral blood RO is a better biomarker than tissue RO in terms of methodological invasiveness and cost. Furthermore, while tissue RO reflects the condition pertaining only to T cells that have already infiltrated the tumor, the peripheral blood RO may reflect the effect of T‐cell activity before T‐cell infiltration. T‐cell activation is also considered to be involved in infiltration from the tumor margins into the interior of the tumor. Therefore, if PD‐1 antibodies are already bound to circulating T cells and these cells reach the tumor margins, the cells may be able to infiltrate the tumor without being suppressed by PD‐L1 signals from the tumor cells, macrophages, or fibroblasts. Indeed, in basal cell or squamous cell carcinomas, the increase in TILs with anti‐PD‐1 therapy is reportedly due to the proliferation of T‐cell clones that have newly invaded the tumor rather than proliferation by reactivation of already existing TILs, and this finding suggests an effect of anti‐PD‐1 therapy on circulating T cells. ³⁴ Therefore, evaluating the RO on T cells in the peripheral circulation may provide an overview of the two steps involved in T‐cell infiltration into cancer tissue and activation at the tumor site, and this process would include immature T cells rather than TILs.

The second limitation is that the small number of cases included a wide variety of carcinomas and histologic types and the prognostic factors for ICI are generally reported to include PD‐L1 expression, MSI‐H, TMB, TILs, gut microbiota, and NLR. The present analysis did not include patients with indications for ICI treatment because of MSI‐H. We performed a multivariate analysis and adjusted for lymphocyte number as a confounding factor that influences the PBMC RO among these ICI‐related prognostic factors. In the sensitivity analysis, which was only performed for adenocarcinomas, the results were significantly similar to those obtained in the main analysis. These results suggest that the RO on eTregs is a prognostic factor in patients receiving ICI treatment. We plan to increase the number of patients and investigate the RO and T‐cell dynamics and prognostic significance in combination with ICI in future studies.

In conclusion, the study results suggest that the RO for PD‐1 on eTregs at 2 weeks after antibody administration may be a prognostic factor in patients receiving nivolumab‐based therapy. The CD4⁺ T_CM fraction in the peripheral blood before treatment is a potential target for PD‐1 inhibitory therapy, suggesting that the inhibition of PD‐1/PD‐L1 signaling on eTregs may attenuate the antitumor effects of the drug.

AUTHOR CONTRIBUTIONS

Masahiro Hosonuma: Conceptualization; data curation; formal analysis; investigation; methodology; project administration; visualization; writing – original draft. Yuya Hirasawa: Data curation; formal analysis. Atsuo Kuramasu: Conceptualization; project administration; supervision; writing – review and editing. Masakazu Murayama: Data curation; formal analysis; investigation. Yoichiro Narikawa: Data curation; formal analysis; investigation. Hitoshi Toyoda: Data curation; formal analysis; investigation. Yuta Baba: Data curation; formal analysis; investigation. Junya Isobe: Data curation; formal analysis; investigation. Eiji Funayama: Data curation; formal analysis; investigation. Kohei Tajima: Data curation; formal analysis; investigation. Midori Shida: Conceptualization; data curation; formal analysis; investigation; methodology. Kazuyuki Hamada: Data curation; writing – review and editing. Toshiaki Tsurui: Data curation; formal analysis; investigation. Hirotsugu Ariizumi: Data curation; writing – review and editing. Tomoyuki Ishiguro: Data curation; writing – review and editing. Risako Suzuki: Data curation; writing – review and editing. Ryotaro Ohkuma: Data curation; writing – review and editing. Yutaro Kubota: Conceptualization; data curation; project administration; writing – review and editing. Atsushi Horiike: Project administration; writing – review and editing. Takehiko Sambe: Project administration; supervision; writing – review and editing. Mayumi Tsuji: Project administration; supervision. Satoshi Wada: Project administration; supervision. Yuji Kiuchi: Funding acquisition; project administration; supervision. Shinichi Kobayashi: Funding acquisition; project administration; supervision. Takuya Tsunoda: Conceptualization; funding acquisition; project administration; supervision. Kiyoshi Yoshimura: Conceptualization; project administration; supervision; writing – review and editing.

CONFLICT OF INTEREST STATEMENT

Kiyoshi Yoshimura has a financial conflict of interest to disclose regarding speaking fees to the following companies. Kyowa Kirin, Bristol Myers Squibb, Chugai Pharmaceutical, MIYARISAN Pharmaceutical. Other authors do not report any conflict of interest.

ETHICS STATEMENT

Approval of the research protocol by an Institutional Reviewer Board: This study was approved by the Ethics Committee of Showa University School of Medicine (Approval #2165).

Informed Consent: Informed consent was obtained from all enrolled patients.

Registry and the Registration No. of the study/trial: N/A.

Animal Studies: N/A.

Supporting information

Figure S1.

Figure S2.

Figure S3.

Figure S4.

Table S1.

Hosonuma M, Hirasawa Y, Kuramasu A, et al. Nivolumab receptor occupancy on effector regulatory T cells predicts clinical benefit. Cancer Sci. 2024;115:752‐762. doi: 10.1111/cas.16061