Advanced Tertiary Lymphoid Tissues in Protocol Biopsies are Associated with Progressive Graft Dysfunction in Kidney Transplant Recipients

Significance Statement

Tertiary lymphoid tissues (TLTs) are frequently found in transplanted kidneys, but their prevalence and clinical significance remain uncertain. Serial protocol kidney transplant biopsies without signs of rejection were collected and TLTs staged according to the presence of proliferating lymphocytes and follicular dendritic cells. TLTs rapidly developed within 1 month after kidney transplantation in approximately half of the 214 patients. Advanced TLTs, defined as the presence of follicular dendritic cells, were associated with progressive decline in graft function independent of interstitial inflammation score. These findings suggest advanced TLTs are strongly associated with late graft dysfunction, even in the absence of rejection.

Visual Abstract

Abstract

Background

Tertiary lymphoid tissues (TLTs) are ectopic lymphoid tissues found in chronically inflamed organs. Although studies have documented TLT formation in transplanted kidneys, the clinical relevance of these TLTs remains controversial. We examined the effects of TLTs on future graft function using our histologic TLT maturity stages and the association between TLTs and Banff pathologic scores. We also analyzed the risk factors for the development of TLTs.

Methods

Serial protocol biopsy samples (0 hour, 1, 6, and 12 months) without rejection were retrospectively analyzed from 214 patients who underwent living donor kidney transplantation. TLTs were defined as lymphocyte aggregates with signs of proliferation and their stages were determined by the absence (stage I) or presence (stage II) of follicular dendritic cells.

Results

Only 4% of patients exhibited TLTs at the 0-hour biopsy. Prevalence increased to almost 50% at the 1-month biopsy, and then slightly further for 12 months. The proportion of advanced stage II TLTs increased gradually, reaching 19% at the 12-month biopsy. Presence of stage II TLTs was associated with higher risk of renal function decline after transplantation compared with patients with no TLT or stage I TLTs. Stage II TLTs were associated with more severe tubulitis and interstitial fibrosis/tubular atrophy at 12 months and predicted poorer graft function independently from the degree of interstitial inflammation. Pretransplantation rituximab treatment dramatically attenuated the development of stage II TLTs.

Conclusions

TLTs are commonly found in clinically stable transplanted kidneys. Advanced stage II TLTs are associated with progressive graft dysfunction, independent of interstitial inflammation.

Kidney transplantation is an ideal treatment for patients with ESKD. Whereas short-term graft survival had been greatly improved in the last three decades, long-term graft survival has changed only marginally.^1,2 Among various factors affecting the outcomes of transplanted kidneys, chronic intragraft inflammation has been considered one of the most important components that contribute to persistent allograft injury.^3–6 Studies have consistently suggested that subclinical rejection frequently leads to deterioration of transplanted kidneys if left untreated.^7–10 Even mild tubulointerstitial inflammation was associated with poor graft outcomes after kidney transplantation, highlighting the relationship between unresolved inflammation and progressive functional decline.¹¹ For these reasons, better understanding and proper management of graft inflammation are prerequisites to long-term graft survival.

Persistent inflammatory stimuli often give rise to the development of tertiary lymphoid tissues (TLTs), namely, inducible ectopic lymphoid tissues that arise in chronic inflammatory conditions such as aging, cancer, autoimmune diseases, and in transplanted organs.^12–15 T and B lymphocytes are the main hematopoietic components of TLTs, and specialized fibroblasts provide structural support and produce homeostatic chemokines, such as CXCL13.^16–22 Although the functional roles of TLTs are context dependent, we have previously described a strong association between renal TLTs and maladaptive repair in rodent models.²⁰

To provide objective and standardized analytic methodology, we recently proposed a new TLT staging strategy on the basis of the presence of follicular dendritic cells (FDCs) and germinal centers, both of which represent cellular components of advanced TLTs.²³ TLT stages positively correlated with the severity of kidney injury and inflammation, suggesting the potential to serve as additional histologic markers of tissue inflammation.

The presence of TLTs in transplanted kidneys is well documented.^24–35 Nevertheless, their clinical relevance remains controversial. The main reasons for these conflicting results include that TLTs were not separated from concurrent rejection, and the definition of TLTs has been inconsistent across studies.^14,36–38 The facts that TLTs were frequently observed both in rejected and tolerated murine allografts further complicate their functional identity in transplanted kidneys.^39–41 To overcome these issues and clarify the effects of TLTs on graft functions, we utilized two major strategies. First, we collected protocol biopsy samples from kidney transplant recipients without overt evidence of rejection to directly investigate the relationship between TLTs and graft function. Second, using our recently established TLT staging method,²³ we classified TLTs on the basis of their phenotypes, and analyzed the association between TLT stages and graft outcomes. Here, we demonstrate that TLTs developed in almost half of patients who are clinically stable and that the presence of advanced stage II TLTs was associated with progressive functional decline of renal allograft, in comparison with stable graft function in patients without TLTs or with stage I TLTs.

Methods

Study Population and Protocol Biopsy Sample Acquisition

An overview of the study design and patient recruitment strategy is given in Figure 1. We retrospectively screened 241 patients who underwent their first living donor kidney transplantation between July 2004 and December 2016 at Akita University in Japan. Four serial protocol biopsies were obtained from each patient during this period. A 0-hour protocol biopsy was performed during cold-saline perfusion after kidney explantation from the donor. Subsequently, recipients underwent protocol biopsies at 1, 6, and 12 months after transplantation. At least two cores of graft tissue were obtained at each biopsy, paraffin embedded, and subjected to conventional histologic stains and immunofluorescence. Patients were excluded if they met one or more of the following criteria: (1) occurrence of biopsy-proven acute rejection within the first year of transplantation; (2) occurrence of BK virus–associated nephropathy within the first year of transplantation; (3) nonrecovery of renal allograft function (<30 ml/min per 1.73 m²) over the first year of transplantation; or (4) loss to follow-up within the first year of transplantation. To be specific, recipients with subclinical rejection (biopsy-proven acute T cell–mediated or antibody-mediated rejection without an elevation in serum creatinine level) were excluded, whereas those with borderline T cell–mediated rejection (t>0 with i0 or i1, or t1 with i2 or i3, without an elevation in serum creatinine levels) were included in this study.^42–44

Figure 1.

A flowchart of the study participant selection. We retrospectively screened 241 patients who underwent living donor kidney transplantation between 2004 and 2016 at Akita University in Japan. After 27 patients with known risk factors for poor graft outcome and those lost to follow-up within a year of kidney transplantation were excluded, the remaining 214 kidney transplant recipients were enrolled. Serial protocol biopsy samples were obtained and processed for immunofluorescence to determine the presence and staging of tertiary lymphoid tissues.

Information regarding the baseline characteristics of the recipients and donors was obtained at the time of kidney transplantation and during visits to the outpatient clinic. Donor-specific antibody was retrospectively measured in the sera 1-year after transplantation stored until use using Luminex-based SAB kits (LABSCreen PRA and LABSCreen Single Antigen, Thermo Fisher Scientific, Waltham, MA). Positive evaluations were made as previously described.⁴⁵ Indication biopsy was performed when patients had an unexplained rise in serum creatinine (>25% from baseline value) during the follow-up period, but the presence of TLTs in these samples was not assessed in this study. Kidney function was measured as the eGFR using the CKD Epidemiology Collaboration formula for the Japanese population.⁴⁶ All diagnoses and Banff pathologic scores were determined and rescored by a single experienced transplant nephrologist, in accordance with Banff 2017 criteria.⁴²

All human specimens were analyzed after informed consent, and the approval of the ethics committees at Akita and Kyoto University hospitals was obtained. This study adhered to the Declaration of Istanbul.

Outcome Measures

The primary end point was the occurrence of death-censored renal function decline, defined as a decline of ≥30% in the eGFR from 1-year post-transplant graft function. The secondary end point was renal allograft function after kidney transplantation.

Immunosuppressive Regimen

Protocols of immunosuppressive drugs are described elsewhere.⁴⁷ Briefly, all patients received basiliximab (20 mg at day 0 and day 4) as induction immunosuppression, followed by maintenance immunosuppression with prednisolone, mycophenolate mofetil, and tacrolimus. Patients undergoing ABO-incompatible kidney transplantation received a single dose of rituximab (200 mg) 3 weeks before transplantation, followed by plasma exchange or double-filtration plasmapheresis and the administration of intravenous Ig. Patients diagnosed with borderline T cell–mediated rejection in protocol biopsies were routinely treated with 10–20 mg/kg of intravenous methylprednisolone for 2–3 consecutive days, depending on the degree of graft inflammation and patient status, unless contraindicated.

Identification and Evaluation of TLTs in Transplanted Kidneys

In this study, we defined TLTs as organized lymphocyte aggregates with the signs of proliferation as described previously.^23,48 Because TLT sizes in transplanted kidneys were variable (Supplemental Figure 1A), we defined organized lymphocyte aggregates as >60 lymphocytes (T cells or B cells) in this study (Supplemental Figure 1, A–C). After diagnosis of TLTs, we determined the stages of each TLT.

Identification and quantification of TLT stages were determined through two steps: (1) identification of mononuclear cell infiltrates in permissive areas for TLT formation, which include subcapsular, periglomerular, and perivascular area, with periodic acid–Schiff (PAS) stained graft sections; and (2) assessment of the lymphocyte infiltrates using immunofluorescence of CD3ε and CD20, and Ki67 and CD21 in two serial sections for each individual, as described previously.²³ After the diagnosis of TLTs, we determined TLT stages of each TLT.

TLT stages were defined as follows:

• 1.TLT lacking either FDC or germinal center: stage I TLT;
• 2.TLT containing FDC but lacking germinal center: stage II TLT; and
• 3.TLT containing both FDC and germinal center: stage III TLT.

FDCs were defined as the cells strongly positive for nonhematopoietic CD21 signals within TLTs. Germinal centers were defined with Ki67-positive cell clusters, which contained >15 Ki67-positive cells per cluster, in B cell areas.

Renal Immunofluorescence

Immunofluorescence studies were performed on tissues from the same block as were used for the preparation of PAS-stained slides. Immunofluorescence staining of biopsy tissues was performed as previously described.⁴⁹ The following primary antibodies were used in these experiments: anti-CD3ε (catalog ab5690; Abcam, Cambridge, UK), anti-CD20 (catalog 14-0202; eBioscience, San Diego, CA), Ki67 (catalog ab16667; Abcam), anti-CD21 (catalog ab75985; Abcam, and catalog MA5-11417; Thermo Fisher Scientific, Waltham, MA), anti-CD45 (catalog 14-9457; eBioscience), anti-p75NTR (catalog AF1157; R&D Systems, Minneapolis, MN), and anti-CXCL13 (catalog AF801; R&D Systems). Staining was visualized using appropriate secondary antibodies. Cell nuclei were counterstained with DAPI. All immunofluorescence samples were analyzed using a confocal microscope (FV1000D; Olympus, Japan).

Statistical Analysis

Statistical analyses were performed using SPSS for Windows, version 20.0 (IBM Corp., Armonk, NY) and STATA 14.1 (StataCorp, College Station, TX). Baseline patient characteristics and clinical parameters were expressed as mean±SD, or as numbers of patients and percentages. Time on dialysis was expressed as median (first and third interquartile range) because it was non-normally distributed. Temporal changes in the TLT stages were assessed by trend analysis. Renal function decline, assessed in a time-to-event analysis, was analyzed by Cox proportional-hazards models with an adjustment for age, sex, the presence of diabetes after transplantation, ABO incompatibility, positive crossmatch, the presence of donor specific antibody at 1 year post-transplantation, and pretransplantation donor eGFR. To further compare the differences in repeatedly measured eGFR during the follow-up period between groups, we used a linear mixed-effect model with robust variance estimation,⁵⁰ adjusting for baseline covariates including donor and recipient age and sex, recipient body mass index, preemptive kidney transplantation, the presence of diabetes after transplantation, the number of HLA mismatching, positive crossmatch, ABO incompatibility, the use of immunosuppressants, baseline graft function, the presence of donor-specific antibody at 1-year post-transplantation, and pretransplantation donor eGFR. Baseline graft function was set as eGFR levels at each biopsy-time point. Logistic regression analysis was performed to identify risk factors for the development of TLTs. The relationship between the use of pretransplantation rituximab and TLTs was analyzed using Pearson’s chi-squared test or Fisher’s exact test as appropriate. Finally, the overall comparisons of Banff pathologic scores with TLT scores were performed using the Kruskal–Wallis test, and the Mann–Whitney test was used for the comparisons of each group. P values <0.05 were considered statistically significant.

Results

Baseline Characteristics of Enrolled Patients

A total of 214 kidney transplant recipients were finally included in this study, and their baseline demographics and laboratory parameters are shown in Table 1. The most common cause of ESKD was chronic glomerulonephritis. Mean eGFRs were 66.3, 65.1, and 62.1 ml/min per 1.73 m² at 1 month, 1 year, and 5 years after kidney transplantation, respectively. Approximately one quarter (53 out of 214, 24.8%) of the enrolled patients underwent ABO-incompatible kidney transplantation; they were older and more frequently received pretransplantation rituximab than those who underwent ABO-compatible kidney transplantation (Supplemental Table 1). Acute rejection accounted for 63.0% (17 out of 27) of the reasons for exclusions in our study; 13 and four patients were attributed to acute antibody-mediated rejection and acute T cell–mediated rejection, respectively (Supplemental Table 2).

Table 1.

Baseline characteristics and clinical parameters of enrolled patients

Characteristics and Parameters
Number of patients	214
Age, yr	48.8±12.5
Sex, male (%)	139 (65.0)
Body mass index (kg/m2)	22.4±3.7
Etiology of ESKD, n (%)
Chronic glomerulonephritis	122 (57.0)
Diabetes mellitus	36 (16.8)
Hypertension	16 (7.5)
Polycystic kidney disease	14 (6.5)
Othera	26 (12.2)
Time on dialysis (month)	18.0 (5.0, 48.8)
Preemptive kidney transplantation, n (%)	40 (18.7)
Number of HLA mismatching (n)	3.3±1.5
Positive crossmatch	14 (6.5)
ABO-incompatible kidney transplantation, n (%)	53 (24.8)
Pre-transplantation rituximab, n (%)	57 (26.6)
Cold ischemic time (minute)	144.0±37.6
Warm ischemic time (minute)	5.0±3.7
Induction immunosuppressant, n (%)
Basiliximab	214 (100)
Maintenance immunosuppressant, n (%)b
Prednisolone	171 (79.9)
Tacrolimus	214 (100)
Mycophenolate mofetil	214 (100)
Borderline T cell–mediated rejection, n (%)c	66 (30.8)
Donor specific antibody at 1-year post-transplantation, n (%)d	13/207 (6.3)
Class I	7 (53.8)
Class II	7 (53.8)
Post-transplant eGFR (ml/min per 1.73 m2)
1 month	66.3±20.6
1 year	65.1±18.9
5 year	62.1±21.6
Donor age, yrs	58.3±9.9
Donor sex, male (%)	82 (38.3)
Pretransplantation donor eGFR (ml/min per 1.73 m2)	103.0±10.8

Data are expressed as mean±SD or the number of patients (percentage). Time on dialysis is non-normally distributed and is expressed as median (first and third interquartile range).

^aOthers include chronic tubulointerstitial nephritis, gestational hypertension, vesicoureteral reflux disease, sepsis, cystinuria, and bone marrow transplant nephropathy.

^bData obtained at the time of outpatient visit 1 year after transplantation.

^cAt least one episode during the first year after transplantation.

^dNot assessed in seven recipients.

Phenotypic Characterization of TLTs in Transplanted Kidneys

PAS-stained graft tissues contained multiple TLT-like mononuclear cell infiltrates (Figure 2A), located in either subcapsular, perivascular, or periglomerular areas (Figure 2, B–D), consistent with our previous study.²³ These infiltrates were composed of T and B cells, (Figure 2, E–G), some of which were proliferating (Figure 2H), meeting our definition of TLTs. Among lymphocyte infiltrates detected in PAS-stained samples, 86.8% were confirmed as TLTs on the basis of immunofluorescence (Figure 2I). In many TLTs, T and B cells were intermingled with one another (Figure 2E), but some TLTs harbored densely packed B cell clusters (Figure 2, F and G). CD21-positive FDCs, that is stromal cells in charge of organizing B cell homeostasis in TLTs,²⁰ were also detected in some B cell clusters (Figure 2, J and K). CXCL13 expression was also observed within TLTs and was colocalized with CD21 but not with CD45 (Figure 2, L and M). Interestingly, T cell–dominant TLTs were detected in the graft tissues of patients treated with pretransplantation rituximab (Supplemental Figure 2, B–D) in permissive areas for TLT formation described above. TLTs in these patients harbored far fewer B cells (Supplemental Figure 2, B–D) than TLTs in aged kidneys and the kidney with CKD (Supplemental Figure 2A),^20,23 but yet had proliferating lymphocytes inside and therefore met the definition of TLTs.

Figure 2.

Characterization of tertiary lymphoid tissues in transplanted kidney. (A–D) Analyses of PAS-stained graft tissues revealing multiple lymphocyte infiltrates in protocol biopsy samples, as indicated by the boxes (A). The clusters were located either (B) under the renal capsule, (C) around blood vessels, or (D) in the periglomerular area. (E–H) Immunofluorescence of (E, F, G) CD3ε (a T cell marker) and CD20 (a B cell marker); and (H) CD45 (a common leukocyte marker) and Ki67 (a proliferation marker). (I) Proportion of tertiary lymphoid tissues among lymphocyte infiltrates found in PAS-stained samples. (J–M) Immunofluorescence of (J) p75 neurotrophin receptor (p75NTR) and CD21; (K) CD21 and CD20; (L) CXCL13 and CD21; and (M) CXCL13 and CD45. Note that p75NTR and CD21 are expressed in FDCs in tertiary lymphoid tissues, and CXCL13 is a main chemoattractant for B cells. Figure 1G shows a magnified view of the white box in Figure 1F. Arrowheads in Figure 1H indicate Ki67-positive proliferating lymphocytes. Scale bars: (A) 1 mm; (B–D) 200 μm; (F) 100 μm; (E, G, J, K) 50 μm; (H, L, M) 10 μm.

The Prevalence and Staging of TLTs in Transplanted Kidneys

We next categorized TLT phenotypes utilizing the TLT staging strategy we recently established (see Methods for details). We observed stage I and stage II TLTs, but not stage III TLTs in protocol biopsies of transplanted kidneys (Figure 3). Notably, the prevalence of TLTs and their stages significantly changed after transplantation (Figure 4). In 0-hour biopsies, TLTs were found in only 3.8% of the samples. This prevalence increased to 46.9% at 1 month after kidney transplantation, and then slightly further within the first year (53.4% and 58.4% in 6- and 12-month biopsies, respectively). By contrast, the development of stage II TLTs was more gradual; their prevalence in 1-month biopsies was comparable to that of 0-hour biopsies (1.4% and 3.6% in 0-hour and 1-month biopsies, respectively) and then began to increase steadily thereafter, reaching 8.6% in the 6-month biopsy samples (6.1-fold increase versus 0 hour) and 18.9% in the 12-month biopsy samples (13.5-fold increase versus 0 hour).

Figure 3.

The staging of tertiary lymphoid tissues in transplanted kidney. Representative immunofluorescence of different TLT stages as determined by the expression patterns of CD3ε, CD20, CD21, and Ki67. Stage I TLTs were defined by the presence of lymphocyte clusters (CD3ε⁺ and CD20⁺) with signs of proliferation (Ki67⁺), and the absence of FDC (CD21^-). Stage II TLTs were defined by the presence of lymphocyte clusters (CD3ε⁺ and CD20⁺) with signs of proliferation (Ki67⁺) along with the presence of FDCs (CD21⁺). Scale bars: 100 μm.

Figure 4.

The prevalence of tertiary lymphoid tissues in transplanted kidney. Relative frequency of patients with no TLT (gray), stage I TLTs (green), and stage II TLTs (orange) at various time points after kidney transplantation. The overall prevalence of TLTs increased from 3.8% at 0-hour baseline to 46.9% at 1 month after kidney transplantation, and then slightly further during 12 months. By contrast, stage II TLTs exhibited a more gradual increase in prevalence, reaching 18.9% at 12 months post-transplantation. *P<0.001 by trends analysis.

Renal Allograft Outcomes in relation to the Presence and Stage of TLTs

Next, we assessed renal allograft functions according to the presence and stages of TLTs at various time points of biopsies. The presence of TLTs, if stages were not taken into consideration, had no significant influence on late graft function (Supplemental Figure 3, A–F). However, when patients were divided according to TLT stage, those with stage II TLTs in the 6- or 12-month biopsies had significantly higher risk of death-censored renal function decline compared with those with no TLT (adjusted hazard ratios of 3.92 and 3.17 at the 6- and 12-month biopsies, 95% confidence interval of 1.23 to 12.47 and 1.25 to 8.02, and P=0.02 and 0.02, respectively; Figure 5, B and C and Table 2). eGFRs over the 5 years after transplantation were also significantly lower in patients with stage II TLTs than in those with no TLT (adjusted mean differences of -14.35 and -11.71 ml/min per 1.73 m² at the 6- and 12-month biopsies, P=0.008 and 0.01, respectively; Figure 5, E and 5F and Table 3). In patients exhibiting stage II TLTs in the 1-month biopsy, the risk of late graft dysfunction was higher and mean eGFR at 1-year post-transplantation was lower than in those without TLTs or in those with stage I TLTs, although without significant difference (Figure 5A, 5D and Tables 2 and 3). Sensitivity analyses of recipients who underwent ABO-compatible kidney transplantation consistently showed the development of stage II TLT in the 6- or 12-month biopsy was associated with significantly higher risk of decline in graft function compared with those without TLTs (adjusted hazard ratios of 3.97 and 2.81 at the 6- and 12-month biopsies, 95% confidence intervals of 1.15 to 13.76 and 1.05 to 7.51, and P=0.03 and 0.04, respectively; Supplemental Figure 4 and Supplemental Tables 3 and 4).

Figure 5.

Renal allograft outcomes according to the staging of tertiary lymphoid tissues. (A–C) The cumulative incidence rate of death-censored renal function decline and (D–F) the longitudinal trends of eGFR after kidney transplantation according to the stages of TLTs at given time points. The labels 1 month, 6 months, and 12 months refer to biopsy time points after kidney transplantation. Patients with stage II TLTs at 6 or 12 months experienced significantly accelerated graft dysfunction compared with those without TLT. The statistical comparisons between groups are performed by (A–C) Cox regression analysis and (D–F) linear mixed-effect models with multiple adjustments, and their results are shown in Table 2 and 3, respectively. (D–F) Data are expressed as mean±SE for each time point of follow-up. **P<0.005, versus no TLT.

Table 2.

Hazard ratios of the stages of tertiary lymphoid tissues for death-censored renal function decline

Biopsy Time Point	TLT Stages	No. of Eventsa (%)	Adjusted HRb (95% CI)	P Value
1 month	No TLT (n=102)	19 (18.6)	Reference	—
	Stage I (n=83)	18 (21.7)	1.44 (0.72 to 2.87)	0.31
	Stage II (n=7)	3 (42.9)	3.60 (0.96 to 13.50)	0.06
6 month	No TLT (n=76)	10 (13.2)	Reference	—
	Stage I (n=73)	15 (20.5)	1.49 (0.63 to 3.53)	0.37
	Stage II (n=14)	6 (42.9)	3.92 (1.23 to 12.47)	0.02
12 month	No TLT (n=77)	11 (14.3)	Reference	—
	Stage I (n=73)	13 (17.8)	1.05 (0.44 to 2.51)	0.91
	Stage II (n=35)	14 (40.0)	3.17 (1.25 to 8.02)	0.02

HR, hazard ratio; 95% CI, 95% confidence interval.

^aRenal function decline was defined as a decline of ≥30% in the eGFR from 1-year post-transplant graft function.

^bThe comparisons between groups are performed by Cox regression analysis with multiple adjustments for confounders including age, sex, the presence of diabetes after transplantation, ABO incompatibility, positive crossmatch, the presence of donor specific antibody at 1-year post-transplantation, and pretransplantation donor eGFR.

Table 3.

Association between the stages of tertiary lymphoid tissues and graft function

Biopsy Time Point	TLT Stages	Adjusted Difference in eGFRa (95% CI)	P Value
1 month	No TLT (n=102)	Reference	—
	Stage I (n=83)	−0.24 (-6.04 to 5.56)	0.94
	Stage II (n=7)	−13.16 (-32.60 to 6.29)	0.19
6 month	No TLT (n=76)	Reference	—
	Stage I (n=73)	−4.07 (-10.07 to 1.93)	0.18
	Stage II (n=14)	−14.35 (-24.93 to -3.76)	0.008
12 month	No TLT (n=77)	Reference	—
	Stage I (n=73)	−1.16 (-5.78 to 3.47)	0.62
	Stage II (n=35)	−11.71 (-20.84 to -2.58)	0.01

95% CI, 95% confidence interval.

^aThe comparisons between groups are performed by linear mixed-effect models with multiple adjustments for confounders including recipient age and sex, donor age and sex, recipient’s body mass index, preemptive kidney transplantation, the presence of diabetes after transplantation, the number of HLA mismatching, positive crossmatch, the use of immunosuppressant, baseline graft function, the presence of donor specific antibody at 1-year post-transplantation, and pretransplantation donor eGFR. Baseline graft function was set as eGFR levels for each time point. The differences in eGFR are calculated by comparing eGFR at baseline and eGFR at 4–5 years after kidney transplantation.

Previous studies suggested a possible association between the presence of TLTs, especially B cell clusters, and the occurrence of alloantibodies and subsequent antibody-mediated injury.^32,33,51 In our cohort, donor-specific antibodies at 1 year after transplantation were more frequently detected in patients with stage II TLTs than in those with no TLTs or stage I TLTs at the 12-month biopsies (Supplemental Figure 5). Nevertheless, no patient was diagnosed with biopsy-proven acute antibody-mediated rejection during 4 years of follow-up after the final protocol biopsy at 12 months. Seven patients had biopsy-proven acute T cell–mediated rejection during this period; however, these episodes were not associated with the prevalence of stage II TLTs at 12-month post-transplantation.

About one third of our patients (66 out of 214) experienced at least one episode of borderline T cell–mediated rejection during the first year after kidney transplantation. Most of the patients were treated with steroid pulse therapy (63 out of 66, 95.5%), and the trends of eGFR between patients with and without borderline rejection were not different (Supplemental Figure 6).

Risk Factors for the Development of Stage II TLTs in the 12-month Biopsies

Logistic regression analysis revealed the use of pretransplantation rituximab was the strong negative risk factor for the development of stage II TLTs in the 12-month biopsies (odds ratio of 0.17, 95% confidence interval of 0.04 to 0.72, P=0.02; Table 4). Pretransplant rituximab administration suppressed stage II TLTs but did not affect the prevalence of stage I TLTs, regardless of biopsy time point (Figure 6 and Supplemental Table 5). The prevalence of stage II TLTs was lower in ABO-incompatible subgroup than in ABO-compatible subgroup, although without statistical significance (Supplemental Table 6). eGFR levels were maintained at similar levels in patients treated with pretransplantation rituximab compared with those who were not, despite ABO incompatibility (Supplemental Figure 7).

Table 4.

Multivariable analyses of risk factors for the development of stage II tertiary lymphoid tissues in 12-month protocol biopsies

Variables	OR (95% CI)	P value
Recipient age (per 10-year increase)	1.08 (0.74 to 1.59)	0.70
Donor age (per 10-year increase)	0.69 (0.38 to 1.24)	0.21
Recipient sex (male)	0.93 (0.30 to 2.85)	0.89
Donor sex (male)	1.84 (0.63 to 5.32)	0.26
Body mass index	0.97 (0.85 to 1.11)	0.68
Diabetes mellitus	1.70 (0.63 to 4.55)	0.29
Number of HLA mismatching (per one mismatch increase)	0.90 (0.65 to 1.24)	0.52
Positive crossmatch	0.86 (0.11 to 7.06)	0.89
Pretransplantation rituximab	0.17 (0.04 to 0.72)	0.02
Steroid maintenance therapy at 1-year post-transplantation	0.55 (0.20 to 1.57)	0.27
Cold ischemic time (per ten-minute increase)	1.02 (0.95 to 1.10)	0.57
Warm ischemic time (per a minute increase)	0.91 (0.76 to 1.09)	0.30
Donor specific antibody at 1-year post-transplantation	7.63 (1.36 to 42.91)	0.02
12-month eGFR (per 10 ml/min per 1.73 m2 increase)	0.96 (0.71 to 1.23)	0.78
Donor eGFR (per 10 ml/min per 1.73 m2 increase)	0.61 (0.35 to 1.06)	0.08
Borderline T cell–mediated rejection	1.88 (0.73 to 4.85)	0.19

ABO incompatibility was not used as variables because of its significant correlation with the use of pre-transplantation rituximab. OR, odds ratio; 95% CI, confidence interval.

Figure 6.

The effects of pre-transplantation rituximab on the prevalence of tertiary lymphoid tissues. Prevalence of (A) stage I and (B) stage II TLTs at each time point of follow-up according to the use of pre-transplantation rituximab. The administration of rituximab before kidney transplantation was associated with a lower prevalence of stage II TLTs, but not with the change in stage I TLTs regardless of biopsy time point.

The presence of donor-specific antibody at 1 year after transplantation was positively associated with 12-month stage II TLTs (odds ratio of 7.63, 95% confidence interval of 1.36 to 42.91, P=0.02; Table 4). Borderline acute T cell–mediated rejection was not associated with the development of either stage I or II TLTs (Table 4 and Supplemental Table 5).

The Association between TLTs and Banff Pathologic Scores

Finally, we investigated the relationship between TLT stages and Banff pathologic scores obtained at 12 months after kidney transplantation (Table 5). The presence of TLTs correlated with more pronounced interstitial inflammation at 12 months, presumably because TLTs themselves are regarded as interstitial inflammation according to the definition of Banff scores.⁴² Nevertheless, small stage II TLTs were occasionally found against the background of trivial Banff i scores (Supplemental Figure 8). Importantly, patients in the stage II TLT group exhibited significantly worse tubulitis, tubular atrophy, and interstitial fibrosis than did patients in the no TLT group (t score of 0.12±0.36 versus 0.60±0.91, ct score of 0.77±0.72 versus 1.29±0.75, and ci score of 0.70±0.63 versus 1.17 versus 0.92; P<0.001, < 0.001, and 0.008, respectively; Table 5). Patients with the stage I TLTs also showed worse tubulointerstitial inflammation, tubular atrophy, and interstitial fibrosis scores compared with those without TLTs (Table 5), although late graft function was similar between these groups (Figure 5F). Moreover, the presence of stage II TLTs at 12-month biopsies might be associated with future graft dysfunction, even in patients with quantitatively mild interstitial inflammation (adjusted hazard ratio of 2.60 and mean eGFR differences of -11.2, 95% confidence interval of 1.01 to 6.70, and P=0.05 and 0.05, respectively; Figure 7 and Table 6). Banff pathologic scores at 12-month biopsy were not significantly different between ABO-compatible and ABO-incompatible subgroups, except for C4d, whose scores were clearly higher in recipients who underwent ABO-incompatible transplantation (Supplemental Table 7).

Table 5.

Banff pathologic score at 12-month post-transplantation by the stages of tertiary lymphoid tissues

12-month Banff Scores	12-month Protocol Biopsy			P Value
	No TLT	Stage I	Stage II	Overall Comparisona	No TLT Versus Stage Ib	No TLT Versus Stage IIb	Stage I Versus Stage IIb
i	0.44±0.64	0.97±0.87	1.06±0.94	0.03	<0.001	<0.001	0.64
t	0.12±0.36	0.38±0.68	0.60±0.91	0.01	0.004	<0.001	0.27
v	0±0	0±0	0±0	1.00	—	—	—
g	0.03±0.16	0±0	0.03±0.29	0.53	—	—	—
ptc	0.08±0.32	0.14±0.39	0.17±0.45	0.39	—	—	—
ct	0.77±0.72	1.01±0.62	1.29±0.75	0.003	0.007	<0.001	0.08
ci	0.70±0.63	0.99±0.68	1.17±0.92	0.05	0.008	0.008	0.39
cv	0±0	0±0	0±0	1.00	—	—	—
cg	0±0	0±0	0±0	1.00	—	—	—
C4d	0.63±1.07	0.84±1.12	0.39±0.76	0.10	—	—	—

Data are expressed as mean±SD. TLTs, tertiary lymphoid tissues; i, interstitial inflammation; t, tubulitis; v, intimal arteritis; g, glomerulitis; ptc, peritubular capillaritis; ct, tubular atrophy; ci, interstitial fibrosis; cv, chronic fibrous intimal thickening; cg, transplant glomerulopathy.

^aKruskal–Wallis test used for overall comparisons.

^bMann–Whitney test was used for between-group comparisons.

Figure 7.

Renal allograft outcomes according to the staging of tertiary lymphoid tissues among patients with mild interstitial inflammation in the 12-month biopsies. (A) The cumulative incidence rate of death-censored renal function decline and (B) eGFR over a period of 5 years after kidney transplantation according to the stages of TLTs in recipients with mild interstitial inflammation (Banff i score of 0 or 1). Patients with stage II TLTs in the 12-month biopsies experienced progressive declines in renal function, although the degree of overall interstitial inflammation was trivial. The statistical comparisons between groups are performed by (A) Cox regression analysis and (B) linear mixed-effect models with multiple adjustments, and their results are shown in Table 6. (B) Data are expressed as mean±SE for each time point of follow-up.

Table 6.

Association between the stages of tertiary lymphoid tissues and cumulative incidence rate of death-censored renal function decline and post-transplant eGFR among patients with mild interstitial inflammation in the 12-month biopsies

TLT Stages at 12 Months	No. of Eventsa (%)	Adjusted HRb (95% CI)	P Value	Adjusted Difference in eGFRc (95% CI)	P Value
No TLT or stage I (n=129)	22 (17.1)	Reference	—	Reference	—
Stage II (n=23)	9 (39.1)	2.60 (1.01–6.70)	0.05	−11.2 (−22.7 to 0.20)	0.05

HR, hazard ratio; 95% CI, 95% confidence interval.

^aRenal function decline was defined as a decline of ≥30% in the eGFR from 1-year post-transplant graft function.

^bThe comparisons between groups are performed by Cox regression analysis with multiple adjustments for confounders including age, sex, the presence of diabetes after transplantation, ABO incompatibility, positive crossmatch, the presence of donor specific antibody at 1-year post-transplantation, and pretransplantation donor eGFR.

^cThe comparisons between groups are performed by linear mixed-effect models with multiple adjustments for confounders including recipient age and sex, donor age and sex, recipient’s body mass index, preemptive kidney transplantation, the presence of diabetes after transplantation, the number of HLA mismatching, positive crossmatch, the use of immunosuppressant, baseline graft function, the presence of donor specific antibody at 1-year post-transplantation, and pretransplantation donor eGFR, baseline graft function was set as 12-month eGFR levels. The differences in eGFR are calculated by comparing eGFR at baseline and eGFR at 4–5 years after kidney transplantation.

Discussion

In this study, we investigated the prevalence and clinical relevance of TLTs in transplanted kidneys without the signs of rejection. We found that TLTs were frequent in rejection-free protocol biopsies and their cellular and molecular phenotypes were similar to those found in aged patients.²⁰ By contrast to stage I TLTs that appeared as early as 1 month after kidney transplantation, stage II TLTs developed gradually over time and were independently associated with progressive graft dysfunction. These data suggest advanced TLTs may help to stratify stable renal allografts without rejection into those with and without risk of functional deterioration.

The presence of lymphocyte clusters in the absence of rejection was first described in heart and lung allografts, in which protocol biopsies are performed more frequently; in these hearts and lungs, the prevalence of lymphocyte clusters ranged from 39% to 58%,^52–57 similar to that of TLTs in our study (Figure 4). The clinical effects of lymphocyte clusters on these allografts have been debated and controversial.^52–57 In the previous study, we showed that TLTs in the kidneys develop through at least three developmental stages irrespective of etiologies, and the developmental progression are associated with the severity of kidney injury in human utilizing surgically resected kidney samples.²³ In this study, utilizing renal biopsy samples, we showed the presence of stage II TLTs was associated with functional decline in graft function, whereas the presence of stage I TLTs was not. These results suggest the presence of FDC, not of B cell infiltrations, is the determinant for future graft dysfunction, and may partly explain the inconsistent results of the clinical significance of graft-infiltrating B cells in transplanted kidneys.^{24–29,32,34,58} We therefore propose evaluating TLTs using our staging strategy, especially focusing on the presence or absence of FDCs. Given FDCs are found in heart allografts,^59,60 application of our TLT staging strategy may clarify hitherto unrevealed functional roles of lymphocyte clusters in other transplanted organs.

Time-dependent changes in the distribution of TLT stages provide valuable information regarding their evolution in renal allografts (Figure 4). Stage I TLTs rapidly developed within 1 month after kidney transplantation. By contrast, the prevalence of stage II TLTs did not change at this time point; rather, it increased in the 6- and 12-month biopsy samples. These findings are consistent with our rodent experimental data, where the proportion of advanced TLTs increased in a time-dependent manner after the injury.²³ Notably, we could not find histologic evidence of stage III TLTs, which had been documented in chronically rejected allograft explanted at >5 years after KT.^26,32,35 We speculate that 12 months were too short for ectopic germinal centers to be established and the intensity of graft inflammation in our patients was substantially lower than that in transplanted kidneys with chronic rejection.

Our TLT staging strategy distinguished between progressors and nonprogressors, even among patients categorized as having Banff i scores of 0 or 1. The paradox of kidney allograft with advanced TLTs being categorized as having minimal or mild interstitial inflammation is explained by differences in the grading systems used to score TLTs and interstitial inflammation in the Banff classification. Because Banff i scores depend on the percentage of the area with inflammatory cell infiltration, biopsy samples with small, but FDC-containing TLTs would be classified as minimal or mild interstitial inflammation (Supplemental Figure 8). Another possible explanation is that interstitial infiltrates in subcapsular cortex and in areas of interstitial fibrosis were included in the assessment of TLTs, but not in the determination of Banff i score. Our data suggest TLTs are inflammatory lesions that are qualitatively different from simple interstitial inflammation and therefore should be assessed in a different manner.

Although the underlying mechanisms were not investigated here, analysis of Banff pathologic scores identified sustained tubular injury as a possible cause of progressive graft dysfunction in patients with stage II TLTs (Table 5). Consistent with this hypothesis, a rodent kidney transplantation model study demonstrated that B cells in TLTs promoted tubulointerstitial fibrosis, possibly by secreting fibrosis-related cytokines.⁶¹ Other studies demonstrated that TLTs were associated with the formation of alloantibodies,^26,62 and the intensity of antibody-mediated alloimmune responses correlated with the maturation status of TLTs,^32,33 suggesting a link between advanced TLTs and antibody-mediated graft injury. In this study, however, no patient showed biopsy-proven antibody-mediated rejection, although donor-specific antibodies were more frequently observed in patients with stage II TLTs (Table 4 and Supplemental Figure 5). Moreover, stage II TLTs were not associated with any pathologic features suggestive of antibody-mediated injury such as glomerulitis (g score), peritubular capillaritis (ptc score), or C4d staining at 12 months post-transplantation (Table 5), consistent with findings of a previous report.⁶³ Taken together, these data suggest stage II TLTs contributed to graft dysfunction presumably via tubular inflammation and fibrosis, at least in the first year after kidney transplantation. The association between advanced TLTs and antibody-mediated rejection should be clarified in further investigations.

The effects of rituximab on the development of TLTs have rarely been investigated. In this study, the administration of pretransplantation rituximab dramatically reduced stage II TLTs ≤1 year after kidney transplantation (Figure 6). These findings are mechanistically reasonable, given the capacity of rituximab to deplete circulating B cells for ≥12 months.⁶⁴ Interestingly, a retrospective study showed that post-transplantation rituximab did not result in the clearance of intragraft TLTs.⁵⁸ Similar findings were consistently reported in other conditions such as autoimmune diseases,^65–71 suggesting the importance of the timing of rituximab infusion for controlling TLT formation. Furthermore, patients treated with pretransplantation rituximab maintained similar graft function over 5 years after kidney transplantation, comparable with those who did not, although the immunologic risk was higher in the ABO-incompatible, rituximab-treated group (Supplemental Figure 7). Nevertheless, it remains uncertain whether this effect was due to the reduction of stage II TLTs or other unrevealed mechanisms of rituximab. It is also possible that plasma exchange and/or intravenous Ig, administered along with rituximab, may be associated with suppression of advanced TLTs.

It is noteworthy that most patients treated with pretransplantation rituximab were ABO-incompatible subgroups (50 out of 57, 87.7%), indicating the significant differences in the prevalence of stage II TLTs between ABO-compatible and ABO-incompatible subgroups might be due to the different baseline demographics and immunologic risks rather than the use of pretransplantation rituximab. It was difficult to investigate the effects of pretransplantation rituximab on stage II TLTs among recipients who underwent ABO-compatible transplantation, because of the small number of rituximab-treated recipients in this population (seven out of 161, 4.3%). Given that pretransplantation rituximab is prescribed exclusively for patients with high immunologic risks, distinguishing the effects of rituximab and immunologic profiles on stage II TLTs would be extremely difficult in the real world.

Older age, an important risk factor for developing TLTs,²⁰ was not associated with their formation in this study (Table 4). One of the reasons for this discrepancy could be that both kidney donor and recipients were much younger compared with those recruited in our previous study²⁰ (i.e., a mean age of 58 [donor] and 48 [recipients] versus 70 years [previous study]). We speculate that sustained inflammatory stimuli from immunologic differences, rather than the age of patients, appear to be a more powerful inducer of TLTs in the setting of kidney transplantation.

A limitation of this study should be mentioned. Approximately 12% (108 out of 856) of biopsy samples were missing in this study, mostly because of refusals by recipients (four [1.9%], 22 [10.3%], 51 [23.8%], and 31 [14.5%] missed samples at 0 hour, 1, 6, and 12 months post-transplantation, respectively); these missingness raises a possibility of unexpected selection bias. Nonetheless, we speculate that the proportion of missing patients were relatively small, considering the difficulties in obtaining serial protocol biopsies from recipients maintaining stable graft function. Furthermore, the baseline characteristics and clinical parameters were mostly comparable between recipients who underwent protocol biopsy and those who did not (data not shown). Therefore, the effects of the missing patients on the overall results might be trivial.

In conclusion, we demonstrated the intragraft detection of stage II TLTs was independently associated with progressive decline in renal allograft function. Prospective studies are needed to confirm whether our novel TLT staging strategy has the potential to serve not only as a valuable tool for systematic classification, but also as a predictor of transplant functional decline. Further investigations are also needed to determine whether therapeutic strategies to prevent the development and maturation of TLTs could lead to better long-term graft outcomes in kidney transplant recipients.

Disclosures

J. Floege reports having consultancy agreements with AstraZeneca, Bayer, Calliditas, Idorsia, Novartis, Omeros, and Travere; and reports receiving honoraria from AstraZeneca, Bayer, Calliditas, Fresenius, Idorsia, Novartis, Omeros, and Travere, and Vifor. M. Yanagita reports receiving research grants from Astellas, Chugai, Daiichi Sankyo, Fujiyakuhin, Kyowa Hakko Kirin, Mitsubishi Tanabe, MSD, Nippon Boehringer Ingelheim, and Torii; reports receiving honoraria from Astellas, Chugai, Kyowa Kirin, and others for lecture honoraria; and reports having other interests/relationships with the International Society of Nephrology and the Japanese Society of Nephrology. S. Fukuma reports having consultancy agreements with Kyowa Hakko Kirin, and Rege Nephro; reports receiving research funding from CANCER SCAN and Kyowa Hakko Kirin; reports receiving honoraria from and being a scientific advisor or member of Kyowa Hakko Kirin. S. Yamamoto reports having other interests/relationships with the Japanese Society of Nephrology, and The Japanese Society for Dialysis Therapy. T. Habuchi reports receiving research funding and honoraria from Astellas Pharma, Bayer, Chugai Pharmaceutical, and Kyowa Kirin. Y. Sato reports being employed by the TMK project. All remai ning authors have nothing to disclose.

Funding

This research was supported by the Japan Agency for Medical Research and Development (AMED) grants 20gm1210009, JP20gm5010002, and JP120gm0610011, TMK Project grants, the Japan Society for the Promotion of Science KAKENHI Grant-in-Aids for Scientific Research B (26293202, 17H04187), Grant in Aid for Scientific Research on Innovative Areas “Stem Cell Aging and Disease” (17H05642) and “Lipoquality” (18H04673), Grant in Aid for Young Scientists (B), the AMED Translational Research program, Strategic Promotion for practical application of INnovative medical Technology, and Uehara Memorial Foundation, Takeda Science Foundation, and the Sumitomo Foundation grants. This work was also partly supported by the MEXT, Japan, grant World Premier International Research Center Initiative. This study was co-financed by the German Research Foundation (SFB/TRR57 and SFB/TRR219, BO3755/3-1, and BO3755/6-1), the German Ministry of Education and Research (STOP-FSGS-01GM1518A), the RWTH Interdisciplinary Centre for Clinical Research (IZKF: O3-7), and the National Research Foundation of Korea (NRF-2021R1G1A1014115).

Supplemental Material

This article contains the following supplemental material online at http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ASN.2021050715/-/DCSupplemental

Supplemental Table 1. Baseline characteristics and clinical parameters of patients according to ABO incompatibility.

Supplemental Table 2. Baseline characteristics, clinical parameters, and pathologic classifications of patients who experienced acute rejection within the first year post-transplantation.

Supplemental Table 3. Hazard ratios of the stages of tertiary lymphoid tissues for death-censored renal function decline in subgroup of recipients who underwent ABO-compatible kidney transplantation.

Supplemental Table 4. Association between the stages of tertiary lymphoid tissues and graft function in subgroup of recipients who underwent ABO-compatible kidney transplantation.

Supplemental Table 5. Multivariable analyses of risk factors for the development of stage I tertiary lymphoid tissues in 12-month biopsies.

Supplemental Table 6. The stages of tertiary lymphoid tissues according to ABO compatibility.

Supplemental Table 7. Banff pathologic score at 12 months post-transplantation according to ABO compatibility.

Supplemental Figure 1. Variations in the sizes of tertiary lymphoid tissues in transplanted kidneys.

Supplemental Figure 2. T cell–dominant tertiary lymphoid tissues in transplanted kidneys treated with pretransplantation rituximab.

Supplemental Figure 3. Renal allograft outcomes according to the presence of tertiary lymphoid tissues.

Supplemental Figure 4. Renal allograft outcomes according to the stages of tertiary lymphoid tissues in subgroup of recipients who underwent ABO-compatible kidney transplantation.

Supplemental Figure 5. The prevalence of donor specific antibody at 12 months after kidney transplantation according to the stages of tertiary lymphoid tissues.

Supplemental Figure 6. Longitudinal trends of renal allograft function according to the history of borderline acute T cell–mediated rejection within the first year after kidney transplantation.

Supplemental Figure 7. Longitudinal trends of renal allograft function according to the use of pretransplantation rituximab and the stages of tertiary lymphoid tissues in the 12-month biopsies.

Supplemental Figure 8. Association between the stages of tertiary lymphoid tissues and interstitial inflammation scores.