PD-1 and PD-L1 Expression in Peripheral CD4/CD8⁺ T Cells Is Restored in the Partial Remission Phase in Type 1 Diabetes

Abstract

Context

Partial remission (PR) in type 1 diabetes (T1D) is accompanied by downregulation of the immune response. Programmed cell death-1 (PD-1) and its ligand (PD-L1) are important immunosuppressive molecules, but their changes in the PR phase are unclear.

Objective

We investigated the dynamic changes of PD-1/PD-L1 expression on T cells around the PR phase in T1D.

Methods

Ninety-eight T1D patients were recruited cross-sectionally and grouped according to PR status into nonremitters (individuals who did not undergo PR during the disease course; n = 39), pre-PR (n = 15), mid-PR (n = 30), and post-PR (n = 14) subgroups. PR was defined according to C-peptide level ≥300 pmol/L or index of insulin-adjusted hemoglobin A1c ≤9 as recommended. Among all the 98 patients, 29 newly diagnosed individuals were prospectively followed up for 1 year. The dynamic changes of PD-1/PD-L1 expression, frequency of regulatory T cells (Tregs) and IL-35⁺ Tregs among peripheral CD4/CD8⁺ T cells were determined.

Results

PD-1/PD-L1 on CD4⁺/CD8⁺ T cells showed a dynamic change around the PR phase: lowest in pre-PR phase, restored in mid-PR phase, and declined again in post-PR phase. Conversely, this pattern did not occur for nonremitters. Notably, PD-1 expression on CD8⁺ T cells in mid-PR was positively correlated with the length of the PR phase. The percentages of circulating Tregs and IL-35⁺ Tregs showed no relation to PR.

Conclusions

The PR phase is associated with restoration of PD-1/PD-L1 on CD4⁺ and CD8⁺ T cells, suggesting that PD-1/PD-L1 may be a potential target for prolonging this phase in T1D.

PD-1, PD-L1, partial remission phase, type 1 diabetes, T cell subsets

Programmed cell death-1 (PD-1) and its ligand (PD-L1) are important immunosuppressive molecules that are mainly expressed on activated T cells or β cells (1). Their association with the pathogenesis of type 1 diabetes (T1D) was recognized long ago. Back in 2003, a study on a nonobese diabetic (NOD) mouse model showed that PD-1/PD-L1 blockade could induce T1D, while PD-1/PD-L1 overexpression could effectively inhibit autoimmune T cell responses and reverse diabetes (2). With the momentous application of PD-1 and PD-L1 antibody therapies in the field of oncology (3), their side effects of evoking autoimmune diabetes (4) have caused substantial concern about the relationship between PD-1/PD-L1 and T1D. Importantly, several studies have shown that patients with T1D have decreased PD-1 expression on T cells (5,6); however, the expression of PD-L1 on β cells could reduce autoimmune attacks and resist T cell-mediated destruction (7), making PD-L1 a potential biomarker for recognizing β cells with the capacity to resist the autoimmune attacks. In this sense, it is of particular interest to explore whether PD-1 and PD-L1 expression on immune cells could be a possible marker for the immune response and even for the progression of β cell function.

As a special stage in T1D, the partial remission (PR) phase is characterized by transient β cell function recovery and downregulation of the immune response (8). Although the PR phase usually lasts no more than a year (9), its presence is associated with better glycemic control and a lower risk of developing chronic complications (10,11). Characterized by short-term restoration of β cell function, patients in PR provide a suitable human cohort for investigating possible markers and potential intervention targets for preserving insulin production in T1D. Recently, downregulation of the immune response, which is characterized by reduced expression of some markers of inflammation or the immune response, was found to accompany the PR phase (12,13). However, the association of PD-1/PD-L1 with the PR phase of T1D remains unclear. Here, we investigated circulating PD-1⁺ and PD-L1⁺ T cells and their changes around the PR phase in T1D and hoped to investigate the possibility of PD-1/PD-L1 as potential biomarkers or targets for prolonging this stage.

Methods

Study design and participants

From February 2017 to December 2018, 520 patients with T1D from a follow-up cohort (to observe the progression of β cell function) at the Second Xiangya Hospital of Central South University (Changsha, Hunan, China) were screened for recruitment (Fig. 1). The diagnosis of T1D was made according to the criteria of the World Health Organization (14), and our inclusion criteria were defined as follows: (i) insulin dependence from the time of disease onset, (ii) positivity for at least 1 of the 3 islet autoantibodies measured (glutamic acid decarboxylase antibody [GADA], insulinoma-associated protein 2 antibody [IA-2A], or zinc transporter 8 antibody [ZnT8A]), (iii) disease duration of less than 2 years, and (iv) regular follow-up to measure C-peptide or insulin dosage and hemoglobin A1c (HbA1c), enabling the determination of PR phase status. Among the 520 patients, 260 were excluded because of long disease duration, and 162 cases were excluded because of irregular follow-up. Ultimately, 98 T1D patients met the inclusion criteria and were enrolled for measurement of PD-1/PD-L1 expression, frequency of regulatory T cells (Tregs) and IL-35⁺ Tregs among peripheral CD4/CD8⁺ T cells.

Figure 1.

Flow chart of patient recruitment.

PR was defined according to the mixed meal tolerance test (MMTT)-stimulated C-peptide levels (≥300 pmol/L) or the index of insulin dose adjusted HbA1c (IDAA1c) (IDAA1C = HbA1c (%) + 4* daily insulin dose/kg ≤9) in the absence of C-peptide values as previously reported (15). A total of 10 patients missed the MMTT test, and their status was defined by IDAA1c. In total, 4 different subgroups were defined according to the PR status: non-remitters, pre-PR, mid-PR, and post-PR. Because PR occurs within 6 months after the disease onset, individuals who did not meet the criteria of PR within 12 months of T1D duration were judged to have no PR during the disease course and were thus defined as nonremitters. Other patients who underwent PR were divided into 3 subgroups: (i) mid-PR patients who met the diagnostic criteria for PR at the time of enrollment and (ii) pre-PR and (iii) post-PR patients, who were defined retrospectively when they did not meet the criteria for PR at the time of enrollment.

The duration of mid-PR was defined from the time the patients fulfilled C-peptide or IDAA1c criteria for the first time (onset of PR) to the last follow-up time (end of PR). Accordingly, pre-PR was determined retrospectively as the last clinical follow-up time before mid-PR, and post-PR was determined as the next clinical follow-up time after the end of PR. The median time for testing PD-1/PD-L1 expression in mid-PR patients was 1 month after the onset of PR, and the median time for testing PD-1/PD-L1 expression in post-PR patients was 2 months after the end of PR. The average time for the pre-PR patients was 0.8 months after T1D onset. The subgroups were depicted in Fig. 1.

Our 98 individuals were enrolled from 2 parts. The first part included 69 individuals who were recruited retrospectively and their PR status was determined by previous follow-up data, and the second part included 29 newly diagnosed individuals who were recruited prospectively for 12-month followed up to determine the PR status. These 69 individuals included nonremitters (n = 25), mid-PR patients (n = 30), and post-PR patients (n = 14). For the other 29 newly diagnosed individuals whose PR status was undetermined, prospective follow-up was performed, of whom 15 patients underwent PR and the other 14 patients were defined as nonremitters. Accordingly, these 29 individuals were grouped into pre-PR (n = 15) and nonremitters (n = 14). So, the total number of nonremitters was 39; with 25 cases were recruited retrospectively and 14, prospectively.

Also, these 29 newly diagnosed patients who were prospectively recruited were followed up for 1 year at 3-month intervals to investigate the dynamic changes in PD-1 and PD-L1 expression. In addition to meeting the previously stated criteria for T1D, patients who were enrolled for the longitudinal follow-up had to fulfill the following criteria: (i) a disease course of less than 3 months, (ii) the mid-PR phase had not been entered yet, and (iii) the PR status could not be determined. In total, 7 patients were excluded because of being in PR at baseline. During the follow-up period, 15 patients underwent PR, and the other 14 patients were nonremitters. One patient dropped out at the ninth month of follow-up. This patient had been observed at the onset and the end of the PR phase prior to drop-out and was therefore included in the analysis. The others had completed at least 4 visits. The time windows for visits were no more than 2 weeks.

Normal controls (NCs) were recruited through a recruitment advertisement at a local preschool education institution and local medical centers. All controls underwent oral glucose tolerance test screening, and we collected their medical history to exclude those with a positive family history of diabetes, infection, autoimmune disease, severe liver or kidney damage, or steroid hormone therapy. We used the method of group matching, and controls were matched for gender, age, height, and weight with T1D patients.

This study was approved by the Ethics Committee of the Second Xiangya Hospital of Central South University, and all subjects provided written informed consent for the study protocol. Physicians recorded body height and weight, waist circumference, hip circumference, and blood pressure. Fasting venous blood samples were tested for triglycerides, total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, fasting plasma glucose, HbA1c, and fasting C-peptide; 2-hour venous blood samples were tested for stimulated serum C-peptide and plasma glucose.

Body mass index (BMI; kg/m²) was calculated as weight/(height squared) in adults. BMIz, BMI was adjusted for age and sex according to the World Health Organization’s child growth standards with the WHO Anthro (0~5 years) and WHO Anthro Plus (6~19 years) programs (16), and used to reflect nutritional status in children and adolescents.

C-peptide and HbA1c assays

HbA1c was measured by automated liquid chromatography (VARIANT II Hemoglobin Testing System; Bio-Rad Laboratories, Hercules, CA, US). Serum C-peptide levels were measured with a chemiluminescence method using an Adiva Centaur Systemakit (Siemens, Munich, Germany). The inter- and intra-assay coefficients of variation were 3.7%-4.1% and 1.0%-3.3%, respectively, as previously reported (17). Peripheral venous blood was drawn at 0, 30, 60, and 120 min after the first bite of mix-meal (MMTT) at baseline. During the 3-month follow-up interval, C-peptide was detected at 0 min only for patients with a C-peptide <16.7 pmol/L at previous visit; C peptide was detected at 0 and 120 min for patients whose C-peptide level was less than 200 pmol/L at previous visit.

Islet autoantibody assays

GADA, IA-2A, and ZnT8A were measured by radioimmunoassay (as previously reported (18-20). Suspected positive samples were tested twice for confirmation, and only the confirmed positive patients were enrolled. The sensitivities of the current GADA, IA-2A, and ZnT8A assays were 78%, 74%, and 70%, respectively. The specificities were 96.7%, 96.7%, and 98.9%, respectively, as evaluated in the Islet Autoantibody Standardization Program (IASP 2012).

Flow cytometry

Fresh venous blood samples from fasting subjects were drawn into sodium heparin tubes and processed within 2 h. Peripheral blood mononuclear cells were isolated by standard Ficoll-Paque Plus density-gradient centrifugation and frozen (-196℃) for analysis at a later time point. T cell subsets were stained with fluorescently labeled monoclonal antibodies (CD3-APC-Cy7, CD4-FITC, CD8-Percp-Cy5.5, PD-1-APC, and PD-L1-PE-Cy7; CD4-Percp-Cy5.5, CD25-BB515, Foxp3-PE-Cy7, Ebi3-APC, and IL12p35-PE; BioLegend, San Diego, CA, US) according to the manufacturer’s protocol. Stained cells were analyzed with a BD FACSCanto II system, which was calibrated daily with appropriate single fluorochrome-stained samples. Fifty thousand lymphocytes were collected using a forward scatter/side scatter (FSC-A/SSC-A) lymphocyte gate and analyzed with FlowJo version 10.0 software (Tree Star, Inc., Ashland, OR, US). Dead cells were excluded from the analysis on the basis of their FSC-A/SSC-A properties and propidium iodide staining. The viability of the thawed samples was 93%±5%. Doublets were excluded by FSC-A/FSC-H. The gating of negative controls was performed by fluorescence minus 1 for PD-1, PD-L1, FOXP3, CD25, EBI3, and IL12p35. An isotype control was used for each panel. The gates of all samples were set so that negative controls uniformly stained <0.5%. Gating strategies for all T cell populations examined, to determine the percentage of CD4⁺/CD8⁺ PD-1⁺/PD-L1⁺ T cells, Tregs (CD4⁺CD25⁺Foxp3⁺), and IL-35⁺Tregs (CD4⁺CD25⁺Foxp3⁺Ebi3⁺IL12p35⁺; Ebi3 and IL12p35 marked IL-35 together), are described in Supplementary Fig. 1 (21).

Statistical analyses

Data are presented as the mean ± standard deviation or as the median (minimum-maximum). Comparisons between 2 groups were made using unpaired Student’s t test. One-way analysis of variance was performed to compare groups after adjustment for potentially confounding variables, including age, gender, and BMI (collinearity analysis showed no obvious confounding among the factors). All variables were mathematically (Kolmogorov-Smirnov test) assessed for normal distribution. A few parameters were not normally distributed, and we opted for nonparametric testing for these variables (Wilcoxon test or Mann-Whitney U-test). Associations of the frequencies of T cell subsets with other parameters were determined by Pearson correlation or Spearman nonparametric correlation. Statistical analyses were performed using SPSS version 25.0 (IBM Corporation, Chicago, IL, US) and GraphPad Prism software version 7 (GraphPad Software, San Diego, CA, US). Differences were considered significant at a 2-tailed P < 0.05.

Results

Decreased PD-1 expression on CD4⁺ and CD8⁺ T cells in patients with T1D

No significant differences in baseline characteristics including age at diagnosis, gender, duration, or C-peptide levels were found between excluded patients (n = 162) and enrolled patients (n = 98) (21). Although BMI was different between the 2 groups, statistical adjustment for BMI did not affect the analysis of PD-1. The clinical features of the enrolled subjects are shown in Table 1. As expected, T1D subjects exhibited significantly lower stimulated C-peptide levels and higher HbA1c. Notably, the frequencies of CD4⁺PD-1⁺ (P < 0.0001) and CD8⁺PD-1⁺ (P = 0.0003) T cells were significantly lower in patients with T1D than in NC subjects. No differences were found regarding the frequencies of CD4⁺PD-L1⁺ T cells, CD8⁺PD-L1⁺ T cells, Tregs, or IL-35⁺ Tregs between the 2 groups.

Table 1.

Clinical features and PD-1/PD-L1 expression in T1D

	NC (n = 49)	T1D (n = 98)	P value
Age (years)	12.0 (1.0-60.0)	12.9 (1.9-65.7)	0.5504
Children and adolescents (＜18 years, %)	67.4	68.4	0.8510
Male (%)	49.0	49.0	1.0000
BMI  0-19 years (BMI z-score)	-0.8 (-5.5-1.8)	-0.7 (-3.6-1.9)	0.5006
>19 years (BMI)	22.9±1.9	18.9±2.7	0.0000
Duration of diabetes (months)	/	4.2 (0-23.7)	/
Stimulated C-peptide (pmol/L)	972.5 (246.4-4115.4)	219.1 (5.5-1315.3)	0.0000
HbA1c (%)	5.1 (4.6-5.6)	8.2 (5.5-16.3)	0.0000
HbA1c (mmol/mol)	32.2 (26.8-37.7)	66.1 (36.6-154.6)	0.0000
DKA onset (%)	/	86.7	/
CD4+ PD-1+ (%)	14.2 (4.7-39.3)	8.7 (1.8-32.1)	0.0000
CD8+PD-1+ (%)	14.3 (1.5-27.9)	9.3 (1.7-25.6)	0.0003
CD4+PD-L1+ (%)	7.0 (0.8-55.1)	6.5 (0.2-32.2)	0.1915
CD8+PD-L1+ (%)	5.3 (1.0-19.5)	3.9 (0.3-16.9)	0.0823
Treg (%)	1.5 (0.8-6.8)	1.5 (0.5-6.4)	0.2815
IL-35+ Treg (%)	63.7 (27.1-87.0)	67.5 (29.6-89.1)	0.4117

Data are presented as mean ± SD or as median (minimum-maximum).

Abbreviations: DKA, diabetic ketoacidosis; NC, normal controls; T1D, type 1 diabetes; Treg, regulatory T cell.

Restoration of PD-1 expression on CD4⁺ T cells in mid-PR patients

To investigate the association of PD-1/PD-L1 expression with PR status, the frequencies of PD-1⁺/PD-L1⁺ T cells, Tregs, and IL-35⁺ Tregs in 30 mid-PR patients were compared with those in patients who were not in the PR phase (including pre-PR, post-PR, and nonremitters combined; n = 68). The results in Table 2 show that PD-1 expression on CD4⁺ T cells was significantly higher in mid-PR patients (P = 0.0065), which was comparable to that in controls (11.45% [3.76%-27.1%] vs. 14.2% [4.72%-39.3%]), indicating that the PR phase was accompanied by restoration of PD-1 expression on T cells. Binary logistic regression analysis confirmed recovery of PD-1 expression on CD4⁺ T cells during the PR phase after adjusting for the confounding effects of age, gender, BMI, and diabetes duration (P = 0.04; data not shown). No differences between these 2 groups were found regarding the frequencies of CD8⁺PD-1⁺, CD4⁺PD-L1⁺, or CD8⁺PD-L1⁺ T cells. The frequencies of Tregs and IL-35⁺ Tregs were also comparable across T1D patients, regardless of PR status.

Table 2.

Clinical features and PD-1/PD-L1 expression in patients during PR phase

	Mid-PR Patients (n = 30)	Patients Who Were Not in the PR Phase (n = 68)	P Value
Age onset (years)	13.0 (5.1-55.2)	12.9 (1.9-65.7)	0.9924
Children and adolescents (＜18 years, %)	76.7	64.7	0.3490
Male (%)	26.7	58.8	0.0040
0-19 years (BMI z-score)	-0.4±1.2	-0.7±1.3	0.3632
>19 years (BMI)	19.6±2.8	18.7±2.7	0.4255
Duration of diabetes (months)	3.1 (0.7-20.6)	6.0 (0-23.7)	0.8944
Stimulated C-peptide (pmol/L)	586.7 (317.6-1315.6)	146.8 (5.5-409.7)	0.0000
HbA1c (%)	7.1 (5.8-14.6)	8.7 (5.5-16.3)	0.0056
HbA1c (mmol/mol)	53.6 (39.9-136.1)	71.6 (36.6-154.6)	0.0055
DKA onset (%)	83.3	88.2	0.2190
CD4+PD-1+ (%)	11.5 (3.8-27.1)	7.7 (1.8-32.1)	0.0065
CD8+PD-1+ (%)	9.6 (2.1-25.6)	9.1 (1.7-25.6)	0.2654
CD4+PD-L1+ (%)	6.9 (1.4-27.3)	5.8 (0.2-32.2)	0.2464
CD8+PD-L1+ (%)	5.8 (0.9-16.9)	3.3 (0.3-16.4)	0.0514
Treg (%)	1.5 (0.8-6.1)	1.6 (0.5-6.4)	0.6093
IL-35+ Treg (%)	67.5 (29.6-86.4)	67.7 (33.7-89.1)	0.6004

Patients who were not in the PR phase (n = 68) included pre-PR (n = 15), post-PR (n = 14), and nonremitters (n = 39). Data are presented as mean ± SD or as median (minimum-maximum).

Abbreviations: DKA, diabetic ketoacidosis; PR, partial remission; T1D, type 1 diabetes; Treg, regulatory T cell.

Dynamic changes of PD-1⁺ and PD-L1⁺ T cells around the PR phase

On the basis of PR status, patients with T1D were divided into pre-PR (n = 15), mid-PR (n = 30), post-PR (n = 14), and nonremitters (n = 39) groups. Notably, the frequencies of PD-1⁺ or PD-L1⁺ CD4⁺ and CD8⁺ T cells decreased in pre-PR patients, significantly increased in mid-PR patients, and then declined to a low level in the post-PR patients (Fig. 2), showing dynamic changes in PD-1⁺ and PD-L1⁺ T cells around the PR phase. However, there were no similar changes in the percentages of Tregs or IL-35⁺ Tregs.

Figure 2.

Changes in PD-1/PD-L1 expression on T cell subsets around the PR phase. The frequency of CD4⁺ PD-1⁺ T cells (A), CD8⁺ PD-1⁺ T cells (B), CD4⁺ PD-L1⁺ T cells (C), and CD8⁺ PD-L1⁺ T cells (D) gated on CD3⁺ CD4⁺ or CD3⁺ CD8⁺ T cells is shown in 4 groups of patients: NC (n = 49), pre-PR (n = 15), mid-PR (n = 30), and post-PR (n = 14). P values refer to comparisons of data after adjusting for age, gender, and BMI. Each point represents the percentage of T cell subsets of an individual. Horizontal lines show medians, 25th and 75th percentiles. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Abbreviations: NC, normal control; PR, partial remission.

To further verify these changes in PD-1 and PD-L1 expression around the PR phase, 29 newly diagnosed T1D patients who had not yet entered the PR phase were followed for 12 months at 3-month intervals. Fifteen patients entered the PR phase, while 14 were nonremitters. Longitudinal testing showed a similar pattern, with PD-1 and PD-L1 expression on CD4⁺ and CD8⁺ T cells increasing significantly in PR patients compared to the nonremitters (Fig. 3).

Figure 3.

Longitudinal changes in PD-1 and PD-L1 expression from disease onset to PR stage. The changes in CD4⁺PD-1⁺ T cells (A), CD8⁺PD-1⁺ T cells (B), CD4⁺PD-L1⁺ T cells (C), and CD8⁺PD-L1⁺ T cells (D) were tested in T1D patients who underwent PR, as well as in the nonremitters. Each line represents an individual, and the dots at each end represent 2 stages (disease onset and PR stage, or the corresponding stage in the nonremitters whose duration was 9.66 ± 2.77 months). Nonremitters are defined as patients who did not undergo PR during the disease course. Abbreviation: PR, partial remission.

Changes in T cells expressing PD-1 and PD-L1 had no association with disease course in the non-remitters

We next analyzed the longitudinal changes in PD-1 and PD-L1 expression on T cells following T1D diagnosis in patients who had PR and in nonremitters. Considering the majority of subjects had their PR within 12 months of disease duration, we included the 83 patients whose PR occurred within 1 year after T1D diagnosis. The other 15 patients had measurements of PD-1/PD-L1 expression at 15th or 21st month, which was slightly far from the mid-PR period, so only 83 patients were included in the analysis. As shown in Fig. 4, the dynamic changes in T cells expressing PD-1 and PD-L1 were seen only in patients who had PR, while no association was found among the nonremitters. This finding indicated that PD-1 and PD-L1 expression in T cells was related to PR status rather than T1D duration.

Figure 4.

Changes in PD-1⁺/PD-L1⁺ T cell percentage based on disease duration. The frequencies of CD4⁺PD-1⁺ T cells (red), CD8⁺PD-1⁺ T cells (green), CD4⁺PD-L1⁺ T cells (blue), CD8⁺PD-L1⁺ T cells (orange), and Treg-cells (black dotted line) based on disease duration were compared between patients who underwent PR (B) and nonremitters who did not undergo PR during the disease course (B). Abbreviation: PR, partial remission.

Positive correlation of PD-1 expression on CD8+ T cells with the duration of PR phase

After adjusting for age, gender, and BMI, the frequencies of CD4⁺PD-1⁺, CD8⁺PD-1⁺, CD4⁺PD-L1⁺, and CD8⁺PD-L1⁺ T cell subsets were not significantly correlated with HbA1c or β cell function (i.e., stimulated C-peptide) during the PR phase. Notably, we found a significant positive correlation between the percentage of CD8⁺ PD-1⁺ T cells during the mid-PR stage and the duration of the PR phase (r² = 0.504, P = 0.0021).

Discussion

T1D is characterized by persistent autoimmune destruction and irreversible failure of β cell function (22). However, the PR or “honeymoon” phase, marked by transient recovery of β cell function, is beneficial for long-term glycemic control. In addition to β cell function recovery, recent evidence has shown that the downregulation of the autoimmune response, as demonstrated by the reduced expression of some inflammatory markers, also plays an important role in the pathogenesis of the PR phase (23). Investigating possible peripheral biomarkers of immune-regulation during PR could contribute to a better understanding of this phase and might help provide potential targets for future immunotherapies aiming to extend the honeymoon period and improve glycemic control following T1D diagnosis. Here, to the best of our knowledge, we first reported that the PR phase of T1D is accompanied by restoration of PD-1/PD-L1 expression on peripheral CD4⁺ and CD8⁺ T cells. These findings suggest that PD-1/PD-L1 might be a biomarker of PR and that this pathway may serve as a potential target for therapies seeking to prolong this stage.

The importance of PD-1/PD-L1 signaling in T1D has long been recognized. PD-1 or PD-L1 blockade accelerated the severity of insulitis and diabetes progression in NOD mice (2); moreover, hematopoietic stem cells expressing high levels of PD-1 effectively inhibited the autoimmune T cell response and reversed diabetes in the NOD mouse model (24). Notably, remission could be induced in the NOD mice model with insulin treatment and that remission was maintained by the PD-1/PD-L1 pathway (25). In human subjects, a few studies have shown that the frequency of CD4⁺PD-1⁺ T cells is low in T1D patients, which is in accord with our results (5,6). However, whether T cell expression of PD-1 and PD-L1 could reflect the immune status of T1D and the PR phase was previously unclear. Our results demonstrated that dynamic changes in PD-1 and PD-L1 expression on CD4⁺/CD8⁺ T cells existed in patients around the PR phase, with expression being the lowest in the pre-PR phase, achieving restoration in the mid-PR phase, and then declining again in the post-PR phase. Granados et al reported that PD-1 expression was decreased on T cells from children with new-onset T1D but was restored 4 to 6 months postdiagnosis (5). Although they did not test C-peptide levels and define PR in their cases, this period of 4 to 6 months postdiagnosis is consistent with the stage of PR. More interesting, PD-1/PD-L1 remained steady during the disease duration for the nonremitters. This finding confirmed the association of PD-1 and PD-L1 with the PR phase rather than T1D duration. Notably, this association was significant only with PD-1 and PD-L1 expression, and no correlation was found with the frequency of Treg or IL35⁺ Treg cells, although increased IL-35 levels were found to be related to the remaining C-peptide levels in long-term T1D patients (26). This might be due to the different immune states in the early and late stages of the disease.

In the cross-sectional results, when we compared mid-PR patients with a group of patients in multiple stages, including pre-PR, post-PR, and no-PR, only PD-1 expression on CD4⁺ T cells was found to be increased significantly. However, both PD-1 and PD-L1 on CD4⁺ and CD8⁺ T cells were increased when the patients were further divided into pre-PR, mid-PR, and post-PR groups. This finding might have gone one step further to support our assumption that the immune status of PR phase was highly varied, which could be partly reflected by PD-1 and PD-L1 expression. There were some slightly different patterns of change in PD-1 and PD-L1 expression in T1D patients in the PR phase. Possible reasons could be related to the differential effects of PD-L1 expression on different cell types. Indeed, one reported study showed that PD-L1 gene expression was upregulated in newly diagnosed T1D patients (27).

There were some concerns that need to be clarified for our patients. Some published studies showed that PR was more frequent with increasing age (28,29), but there was no difference in age between the PR patients and nonremitters in our study. This inconsistency might be due to differences in enrolled subjects because most studies focused on children and adolescents, but our study included adults as well. Additionally, the proportion of diabetic ketoacidosis (DKA) at onset was quite high in our cases. As a Chinese study focusing on the coverage, cost, and care of T1D (3C study) showed, the proportion of diabetic ketosis and DKA was 92.9% in juvenile-onset patients and 83.8% in adults (30). This might be due to the rarity of T1D and lack of disease awareness in China. To eliminate the confounding influence, we studied the effect of DKA and age on PD-1/PD-L1 expression, but no correlation was found. Additionally, there is some possibility that “fast remitters” are not included in the longitudinal observation, because patients who were already in mid-PR phase at baseline were excluded.

Given the immunosuppressive function of PD-1/PD-L1 signaling, it is possible that PD-1 and PD-L1 may promote immune tolerance in the PR phase (31,32), which might lead to the recovery of β cell function. However, we did not find any factors related to increased PD-1 expression in the PR phase, including C-peptide and HbA1c. In previous studies, IL35⁺ Treg frequencies were positively correlated with residual C-peptide secretion in long-standing T1D patients (12,26,33). Furthermore, our study did not show a similar association of IL35⁺ Tregs with C-peptide levels in patients during the PR phase. These discrepancies might be a result of the relatively small sample size and different patient populations studied. A longitudinal study with a larger sample size and more extensive immunophenotyping should be performed to illustrate the exact mechanisms.

Interestingly, we report a positive correlation between CD8⁺PD-1⁺ T cells and the length of the PR phase. The mechanisms governing this association are not yet clear and warrant further investigation. Altogether, our data support the notion that immune mechanisms are involved in β cell function during the PR phase. More studies, especially those evaluating immune and islet cell function, are required.

Considering the broad utility of anti-PD-1/PDL-1 monoclonal antibodies in the field of cancer, their immune-related adverse events, especially checkpoint inhibitor-associated insulin-dependent diabetes mellitus, should be noted (34,35). In this sense, PD-1 should be considered a promising marker for immune intervention for prolonging immunoregulation and the PR phase of T1D, which would be beneficial for improving the long-term adverse outcomes of T1D patients (36). We found that the PR phase of T1D is accompanied by the restoration of PD-1 and PD-L1 expression on peripheral CD4⁺ and CD8⁺ T cells. Although the specific mechanisms are unclear, these results strongly imply that the change in PD-1⁺ and PD-L1⁺ T cell subtypes is related to the downregulation of the autoimmune response in PR (37). Therefore, it is promising to consider the possibility of PD-1 or PD-L1 as a biomarker and potential target for prolonging this stage. While there are trials focusing on Treg intervention in T1D, the transfusion of ex vivo expanded autologous PD-1⁺ and/or PD-L1⁺ T cells might provide a new and effective therapeutic strategy for T1D.

Acknowledgments

The authors thank Paolo Pozzilli (Institute of Endocrinology & Metabolic Diseases at the University Campus Bio-Medico, Rome, Italy) and Drs. Amanda Posgai and Mark Atkinson (University of Florida) for critically reading and editing the manuscript. Parts of this study were presented as a poster at the 79th American Diabetes Association Scientific Sessions, San Francisco, CA, US, June 7-11, 2019.

Funding: This study was supported by the National Key R&D Program of China (Grant NO. 2017YFC1309604, 2016YFC1305000), the Science and Technology Major Project of Hunan Province (Grant NO. 2017SK1020) and the Natural Science Foundation of Hunan Province, China (Grant NO. 2019JJ40419).

Clinical Trials.gov ID: NCT03610984.

Author Contributions: X.L. designed the study and wrote the manuscript. T.Z. did the research, analyzed the data, and wrote the manuscript. X.L. and Z.Z. reviewed and edited the manuscript. R.T. and C.W. contributed to the experiments. Y.X., F.L., and J.L. contributed to the data collection. X.L. and Z.Z. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Additional Information:

Disclosure Summary: The authors have nothing to disclose and no potential conflicts of interest relevant to this article were reported.

Data Availability: The data sets generated and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.