2.1 Patient population

We conducted a retrospective chart review at two tertiary pediatric kidney transplant centers – BC Children's Hospital (BCCH) in Canada and University of Alabama Birmingham (UAB) in the United States. Both centers routinely perform SB, indication biopsies (IB), and FUB to evaluate response to rejection treatment as standard of care. Patients were included in the study if aged 1–20 years old at transplant, transplanted between January 2008 and December 2020 with at least one FUB after a first episode of biopsy-proven rejection episode in the first 24 months post-transplant (referred to as the index biopsy). Index biopsies were either SB or IB. Patients were excluded if they had BK viremia at the time of the index biopsy with rejection, or if there was evidence of antibody-mediated rejection (AMR), since treatment for these entities would be different from treatment for pure TCMR that was being evaluated in this study.

At BCCH, all participants had separately consented and assented (where applicable) to the prospective transplant data registry at the center (REB #H16-01140), and this analysis was approved by the hospital research ethics board (REB #H20-01502) with a waiver of additional informed consent/assent. At UAB, the research activities for this study were approved under IRB protocols #30006822 and #170428001. Study procedures at both sites adhered to the guidelines set forth in the Declaration of Helsinki.

The standard immunosuppression at both sites consisted of basiliximab and methylprednisolone induction, and initial maintenance with tacrolimus (TAC), mycophenolate mofetil (MMF), and prednisone. At BCCH, drug targets for tacrolimus were 10–15 ng/mL through week 4, followed by 8–12 ng/mL through week 12, 6–10 ng/mL through week 26, and 5–8 ng/mL thereafter. Starting MMF dosage was 1200 mg/m²/day and then titrated according to pharmacokinetic studies at month one to achieve an area-under-curve exposure between 30–60 mg h/L. At UAB, tacrolimus targets were 10–12 ng/mL through week 4, followed by 8–10 ng/mL through week 12, and 5–8 ng/mL thereafter. The starting MMF dosage was 1200 mg/m²/day for 48 h than 900 mg/m²/day thereafter, with further adjustments based on clinical evidence of medication adverse effects.

2.2 Allograft kidney biopsies

SB were planned at 1.5, 3, 6, and 12-months post-transplant at BCCH. They were planned at 3 and/or 6 months at UAB. IB included in this study were performed to investigate increased creatinine.

All biopsies were performed under ultrasound guidance using 16- or 18-gauge spring-loaded biopsy needles. A minimum of two tissue cores were obtained for histological analysis and fixed in 10% neutral buffered formalin. Three-micrometer paraffin-embedded serial sections (18 levels per biopsy) were cut and representative sections stained with hematoxylin–eosin, periodic acid Schiff, periodic acid silver methenamine and Masson's trichrome. For suspected glomerular disease, a portion of cortex was submitted for direct immunofluorescence staining (IgG, IgM, IgA, C3, C1q, fibrinogen, kappa and lambda) and electron microscopy. Complement-derived C4d deposition was assessed using indirect immunofluorescence (on frozen tissue) or immunohistochemistry (on formalin fixed paraffin-embedded tissue) on all biopsies. Immunohistochemistry for SV-40 was used to identify BKV infection.

Biopsies were graded by renal pathologists at the respective sites according to the most recent Banff criteria,^{8, 16} and adhered to adequacy requirements (minimum 10 glomeruli and 2 cores of cortex). Borderline TCMR was defined according to the Banff 2018 reference guide as a minimum score of i1t1.¹⁶ Isolated tubulitis from earlier versions of Banff guidelines were considered as normal/stable biopsies.¹⁷ Acute TCMR was, therefore, defined as a score of ≥i1t1, and sub-categorized as borderline and grade ≥IA (≥i2t2).

Clinical data collected on the day of the index biopsy included creatinine, height, treatment changes for rejection, TAC levels, and MMF dose, where data was available. Estimated glomerular filtration rate (eGFR) was calculated by the Schwartz equation¹⁸ using creatinine and height collected from the day of biopsy.

2.3 Clinical management of rejection

Clinical management was recorded based on the findings at index biopsy. Treatment for index rejection episodes and persistent rejection on FUB was prescribed at the discretion of the treating nephrologist at the local center. Patients with biopsy-proven TCMR were typically treated with high-dose IV methylprednisolone (300 mg/m²/dose, maximum of 1000 mg per dose) for three doses, followed by a prednisone taper. This is an accordance with the KDIGO guidelines.¹⁹ Depending on the clinical context and severity of rejection, some patients were also treated with antithymocyte globulin (ATG; 5–7.5 mg/kg/course divided over multiple doses). Some patients also had adjustment in their baseline immunosuppression (i.e., increased TAC, MMF, and/or prednisone doses); however, these data were not collected from the UAB cohort and, therefore, not included in the analyses.

2.4 Follow-up biopsies

FUB were planned for 1.5–2 months after each index biopsy with rejection and were scored in the same manner as detailed above. Clinical data collected on the day of FUB included serum creatinine, height, viral reactivation, TAC levels, MMF dose, and blood PCR for CMV, EBV, and BKV.

2.5 Primary and secondary outcomes

The primary outcome was the extent of rejection resolution at FUB, categorized as a binary outcome of complete resolution of rejection (defined as <i1t1) vs. incomplete resolution (defined as anything ≥i1t1). Secondary outcomes included evaluation of immunosuppressant drug exposure with treatment response, viral complications, and allograft function after the initial FUB. Immunosuppressant exposure was evaluated based on TAC levels or MMF dosage on the day of FUB and association with the primary outcome. Viral reactivation, a potential consequence of increased immunosuppression for treatment of rejection, was identified for any viral load becoming detectable after the index biopsy until the FUB. This included viral load for CMV, EBV, and BKV on the day of FUB. The functional outcome after treatment of rejection was evaluated based on the change of eGFR from the date of the FUB, to the best eGFR after 6 months post-FUB (range 6–12 months) and after 12 months (range 12–18 months), where data was available. Graft survival was also reported, stratified by the primary outcome to explore the influence of rejection resolution on allograft failure.

2.6 Statistical analysis

Normally distributed data are summarized as mean ± standard deviation, non-normal data as median (interquartile range), and counts with proportion (%). Comparisons between groups were performed using Chi-squared analysis for categorical variables and one-way analysis of variance (ANOVA) or student's t test for continuous variables, depending on the number of subgroups in the analysis. Statistical analysis was performed using R, with p <.05 considered to be statistically significant.²⁰

3.1 Study demographics

Fifty-nine patients transplanted in the programs between January 2008 and December 2020 met inclusion criteria of at least one FUB after a first biopsy-proven rejection episode in the first 24 months post-transplant. Of those patients, 32 were from BCCH and 27 were from UAB. One patient from BCCH was excluded from the analysis for BKV viremia at time of index biopsy, leaving 58 patients for the final analysis. Complete data on the primary outcome was available in all cases. The study cohort reflects a typical pediatric kidney transplant population^{2, 5, 21} that is predominately male sex (n = 35, 60%) and young adolescent age at transplant (mean 11.7 ± 5.7 years) (Table 1). There was a higher percentage of Black patients (52%) at UAB compared to BCCH (7%). The majority of transplants were from deceased donors (65% at BCCH and 85% at UAB) and were first-time transplants (91%). The median number of HLA class I and II mismatches were 5/6 (IQR 3, 5) and 2/4 (IQR 2, 3), respectively. There were no patients in either cohort that had pretransplant donor-specific antibodies (DSA) present. Seven patients at UAB developed DSA at the time of index rejection episodes, while no patients at BCCH had DSA formation. A total of 11 patients (19%) experienced graft loss, at a mean of 20.9 ± 8.6 months post-transplant. There were two patient deaths in the cohort; one was from trauma and the other was an unknown outside of hospital death. The average time of follow-up in patients who did not experience graft loss or death was 26.9 ± 9.5 months post-transplant.

3.2 Index rejection episodes

Severity of initial rejection episode was summarized as borderline (n = 34, 59%) and grade ≥1A TCMR (n = 24, 41%) (Table 2). The mean time post-transplant to the initial rejection episode was 6.1 ± 4.7 months (6.7 ± 4.7 months in BCCH, 5.4 ± 4.6 in UAB, p-value = .13), with no difference in time to first rejection episode based on rejection severity (6.1 ± 4.6 months in borderline, 6.1 ± 5.0 months in grade ≥1A; p-value = .48). The index rejection episodes in the cohort were detected more commonly on SB (64%) than IB (36%), with a greater proportion of borderline TCMR detected on SB compared to grade ≥1A TCMR (77% vs. 46%, p = .017). BCCH had more index rejection episodes detected on SB than IB (90% in BCCH vs. 33% in UAB, p-value <.001), and more borderline TCMR than grade ≥1A (74% in BCCH vs. 40% in UAB, p-value = .004). Among the 58 rejection episodes, 53 received treatment for rejection, with the five patients who did not receive rejection treatment reporting borderline TCMR. Treatment with IV steroids was administered in 52 cases of rejection (90%) and five patients were additionally treated with ATG, four of which had grade ≥ 1A rejection. The one case of borderline TCMR that was treated with ATG also had vascular inflammation and severe allograft dysfunction; therefore, ATG was administered for the vascular involvement despite mild tubulointerstitial inflammation. The remaining case had borderline TCMR and was managed with increased maintenance immunosuppression without additional treatment. The eGFR on day of index rejection biopsy was numerically lower among patients with grade ≥1A TCMR versus borderline TCMR; however, the difference was not statistically significant (61 ± 27 vs. 74 ± 32 mL/min/1.73 m², p-value = .081).

3.3 Functional monitoring was uninformative for identifying rejection severity or persistence

On average, the change in eGFR from the day of index biopsy to FUB was variable, with no statistical difference between the rejection subgroups (p-value = .26). Stratified by severity of the index biopsy, the mean eGFR on day of FUB was 72 ± 26 mL/min/1.73 m² in the borderline TCMR group, and 62 ± 19 mL/min/1.73 m² in the grade ≥ 1A TCMR group (p-value = .082). When grouped by complete versus incomplete resolution of rejection on FUB, those with complete resolution had higher eGFR on day of FUB compared to those with incomplete resolution, but the difference was not statistically significant (74 ± 27 vs. 64 ± 20 mL/min/1.73 m²; p-value = .067) (Table 4).

3.4 Follow-up biopsy identified a large proportion with persistent rejection

Fifty-eight FUB were done at a mean of 1.9 ± 0.6 months post-index biopsy in the borderline TCMR group, and at 1.4 ± 0.6 months in the grade ≥1A TCMR rejection group (p-value <.001) (Table 3). Acute rejection resolved (Banff acuity score < i1t1) in 26 (45%) but was persistent in 32 (55%), including borderline TCMR in 16 (28%), and grade ≥1A TCMR in 16 (28%). Those with borderline TCMR on index biopsies had higher rates of complete resolution of rejection on FUB compared to those with grade ≥1A TCMR rejection (53% vs. 33%; p-value = .033).

In the borderline TCMR group that had persistent rejection on FUB (n = 16), 11 patients had ongoing borderline (32%), but five had worsening to grade ≥1A TCMR (15%). Among the five with worsening of rejection, four received initial treatment with IV steroids and one was not initially treated for rejection. In the grade ≥1A TCMR group that had persistent rejection (n = 16), five had improvement to borderline TCMR (21%), whereas 11 had persistent grade ≥1A TCMR (46%) despite standard treatment. Among those that were treated initially with ATG in addition to steroids (n = 5), the FUB demonstrated complete resolution in three patients, while two patients had persistence of grade ≥1A TCMR.

Twelve (24%) patients had experienced viral reactivation by the time of FUB, with a trend for more reactivation in those with incomplete resolution of rejection (32% vs. 13%, p-value = .057). There were five cases of CMV, four cases of EBV, and five cases of BKV; two patients had simultaneous reactivation of 2 viruses.

3.5 Allograft and functional outcomes after the FUB

Incomplete resolution of rejection was further treated in 25 cases (78%), generally with pulse steroids with or without addition of ATG along with an increase in maintenance immunosuppression; usually until resolution was histologically confirmed. Of the seven untreated FUB rejection episodes, five were borderline TCMR and two were grade ≥ 1A TCMR rejection. There was a trend toward improved eGFR at 6–12 months post-FUB, with average eGFR of 61 ± 28 and 77 ± 32 mL/min/1.73 m² among those with incomplete resolution and complete resolution, respectively (p-value = .071). However, this difference was not reliably seen at 12–18 months post-FUB, with average eGFR of 60 ± 27 and 69 ± 29 mL/min/1.73 m² among those with incomplete resolution and complete resolution, respectively (p-value = .30).

Relative to the eGFR at the time of the FUB, there was minimal increase in eGFR at 6–12 months post-FUB (0.4 ± 25.9 mL/min/1.73 m²), and no significant difference in the change in eGFR between the complete resolution and incomplete resolution groups (3.2 ± 31.9 and −1.8 ± 20.4 mL/min/1.73 m², respectively; p-value = .26) (Table 4). The mean change in eGFR at 12–18 months was −6.2 ± 25.2 mL/min/1.73 m² (−6.9 ± 30.4 mL/min/1.73 m² in complete resolution vs. −5.6 ± 19.6 mL/min/1.73 m² in incomplete resolution, p-value = .43).

Eleven patients experienced graft loss (19%), at an average of 20.9 months post-transplant, of which four (15%) had complete resolution and seven (22%) had incomplete resolution (p-value = .27). No association was found between graft loss and the primary outcome (p-value = .49).

3.6 Impact of treatment variability on outcome

We tested for association between type of treatment for rejection and the binary primary outcome of complete versus incomplete resolution of rejection. We found no significant differences between treatment for rejection (i.e., none, IV steroids, IV steroids and ATG) and resolution of rejection (p-value = .56); however, the analysis was not stratified by the severity of the initial rejection episode. There was no significant association between the TAC level (8.2 ± 2.5 ng/ml vs. 7.7 ± 3.3 ng/ml, p-value = .59) or MMF dosage (628 ±246 mg/m2/day vs. 597 ± 222 mg/m2/day, p-value = .33) on day of the FUB biopsy with the primary outcome.