Urinary Cell Gene Signature of Acute Rejection in Kidney Allografts

Kidney allograft recipients

We measured urinary cell levels of CD3E mRNA, IP-10 (CXCL10) mRNA and 18S rRNA in 145 urine samples matched to 145 kidney allograft biopsies from 126 unique kidney allograft recipients. The study was approved by the Weill Cornell Medicine Institutional Review Board and study participants provided written informed consent for participation. The investigation was performed in compliance with the ethical principles specified in the Declaration of Helsinki.^21–22

Kidney allograft biopsies

Percutaneous needle core biopsies were performed under ultrasound guidance at the NewYork Presbyterian Hospital/Weill Cornell Medicine. Biopsies were fixed and stained using hematoxylin eosin, periodic acid Schiff and Masson trichrome stains. Biopsies were classified and graded by our transplant pathologists (S.S and S.V.S) using the Banff 2019 update of the Banff ‘97 classification of allograft pathology.⁴

Biopsies were stained for complement component 4 degradation product (C4d) and Simian Vacuolating virus large T antigen (SV40 large T antigen). Intragraft presence or absence of C4d was determined by immunofluorescence microscopy of frozen section. The presence or absence of SV40 large T antigen was determined by immunostaining with mouse monoclonal IgG2a antibody PAb416 reacting with both SV40 antigen and BKV large T antigen.

Urine biospecimens, isolation of RNA and reverse transcription

Approximately 50 mL of urine was obtained from each recipient using a sterile urine cup at the time of biopsy and was processed for mRNA profiling at our Gene Expression Monitoring (GEM) Core at Weill Cornell Medicine, New York, NY. Urine samples were centrifuged at 2000g for 30 minutes at room temperature to prepare urinary cell pellets. RNalater (50µl), RLT buffer and Beta-mercaptoethanol (350 μl) were added to the sedimented cells and lysed. Total RNA was isolated from the cell lysate using the RNeasy mini kit (Qiagen, Cat. 74104). RNA purity (ratio of A260/A280) and yield (absorbance at A260) were measured using NanoDrop ND-1000 spectrophotometer (ThermoFisher Scientific).

Total RNA was reverse transcribed (RT) to cDNA using TaqMan reverse transcription kit (Cat. N808-0234, Applied Biosystems). RT was performed at the total RNA concentration of 1.0 μg in 100 μl volume and the RT reaction contained 1x TaqMan RT buffer, 500 μM each of 4 dNTPs, 2.5 μM of Random Hexamer, 0.4 Unit/μl of RNase inhibitor, 1.25 Unit/μl of MultiScribe Reverse Transcriptase and 5.5 mM of Magnesium Chloride. The RT mixture was incubated at 25°C for 10 min, 48°C for 30 min, and 95°C for 5 min.

Design and development of gene-specific oligonucleotide primer pairs and TaqMan probes

We designed gene-specific oligonucleotide primers and TaqMan fluorogenic probes (hydrolysis probes) for the measurement of mRNA levels for CD3E, IP-10, TGF-β1 and 18S rRNA; TGF-β1 mRNA was measured as a QC parameter of customized RT-qPCR assays and 18S rRNA served both as a component of the CTOT-04 signature and as a QC parameter. The sequence and location of the gene-specific oligonucleotide primer pairs and TaqMan probes utilized have been reported.¹⁸

Preamplification of cDNA with gene-specific primer pairs

We developed the preamplification protocol to compensate for low yield of total RNA from urinary cells and to measure multiple mRNAs from small amounts of cDNA.¹⁸ The preamplification reaction for each sample was set up to a final reaction volume of 10.0 μl containing 3.0 μl cDNA (from RT of 1 μg total RNA in 100 μl buffer), 5.0 µl Platinum® Multiplex PCR Master Mix, 1.68 µl primer mix (50 μM sense and 50 μM antisense primer) and 0.32 μl water to a final volume of 10 μl. The PCR was set up in a Veriti thermal cycler (Applied Biosystems) and the reaction profile consisted of an initial hold at 95°C for 2 min, 11 cycles of denaturing at 95°C for 30 seconds, primer annealing at 60°C for 90 seconds, extension at 72°C for 1 min, and final hold at 72°C for 10 min. At the end of 11 cycles of amplification, 290 μl of TE buffer was added to the PCR reaction, and 2.5 μl of diluted PCR amplicons were then used for quantification of mRNA using the RT-qPCR assay.

Absolute Quantification of mRNA in customized RT-qPCR assay

We quantified absolute levels of mRNAs using in-house synthesized BAK amplicon in the customized RT-qPCR assay. The Bak amplicon was diluted to a concentration of 1E+7copies/μl (stock solution) and the stock solution was diluted over 6 orders of magnitude (1,000,000, 100,000, 10,000, 1,000, 100, and 10 copies per 1 μl) (work solution) to develop the universal standard curve. The standard curve was developed by adding 2.5 μl of diluted stock solution to duplicate wells and amplified with Bak-specific primer pair and Bak-specific fluorogenic TaqMan probe. The threshold cycles (C_T) were plotted against the log of the initial amount of the Bak amplicon to develop the standard curve. The standard curve copy numbers in our RT-qPCR assays ranged from 25 to 2.5 million copies.

Calculation of CTOT-04 signature

CTOT-04 signature in urinary cells was calculated using the previously reported equation: – 6.1487+0.8534log₁₀(CD3E/18S) + 0.6376log₁₀(IP-10/18S) +1.6464log₁₀(18S).¹⁸

Data analysis

Log₁₀ transformation of mRNA copy number reduced deviation of mRNA copy number from the normal distribution. mRNA copy number in each urine sample was normalized by calculating the ratio of mRNA copy number to copy number of the reference gene 18S rRNA in the same urine sample. mRNA copy numbers <25 (including 0) were scored as 12.5 copies per microgram of total RNA before 18S rRNA normalization. T test was used for two group comparisons, one-way Analysis of Variance (ANOVA) for multiple group comparison and Dunnett’s test following ANOVA for pair-wise comparisons. The ability of the CTOT-04 signature to discriminate between Banff biopsy diagnostic categories was evaluated using receiver operating characteristic (ROC) curves and area under the ROC curve (AUC) and 95% confidence intervals (95% CI) were calculated. For robust statistical inference, we generated 5000 bootstrap sample sets and estimated the optimism-corrected AUCs of ROC curves and bias-corrected 95% CI. Sensitivity and specificity were determined using the previously identified CTOT-04 signature cutpoint of -1.213, the Youden index that maximized sensitivity and specificity for discrimination of those with TCMR biopsies from those with No Rejection (NR) biopsies.¹⁸ The associations of the CTOT-04 signature with the covariates eGFR and time since transplantation to biopsy were analyzed using the R function lm() from the “stats” package (R version 4.2.1) in univariate and multivariate models,²³ and associations with the biopsy diagnosis were computed using the R function glm(), e.g., glm(as.factor(NRvs.AR)∼CTOTsig+eGFR,=“binomial”).²⁴ The corresponding summary statistics were obtained using the corresponding R summary.

Characteristics of Kidney allograft recipients and biopsy-related parameters

The study cohort consisted of 126 unique kidney allograft recipients (patients) and 145 kidney allograft biopsy-matched urine specimens. The median (25^th percentile and 75^th percentile) age was 48.3 years (36.0 and 48.3) and there were 56 females and 70 males. The eGFR, calculated using the 2021 CKD-EPI equation without using race²⁵ was 27 ml (16 and 30), and time since transplantation to biopsy was 7.3 months (1.8 and 39.8). Table 1 shows patient characteristics stratified by biopsy diagnosis. Biopsies were selected from our Biorepository to represent all three major types of AR. As reflected by the higher number of biopsies (n=145) compared to patients (n=126), 17 patients contributed more than one biopsy, and among those biopsies from 6 patients belonged to more than one diagnostic category. As the control group for patients with AR biopsies, we selected patients with biopsies with no rejection features and varying degrees of tubular injury (NR).

Kidney allograft biopsy-related parameters, stratified by biopsy diagnosis, are shown (Table 2). Among the 145 biopsies, 138 were clinically indicated (for-cause) biopsies. All biopsies were judged adequate by pathologists and the mean (SD) number of glomeruli in each diagnostic category met or exceeded Banff thresholds for adequacy (Table 2). Ninety-five of the 145 biopsies were grouped as AR biopsies with 47 biopsies from 42 patients categorized as TCMR; 28 biopsies from 24 patients classified as ABMR; and 20 biopsies from 18 patients classified as MR. Fifty biopsies from 48 patients were classified as NR. All biopsies were negative for intragraft SV40 large T antigen.

CTOT-04 signature score is diagnostic of acute rejection in kidney allografts

Figure 1 displays the CTOT-04 signature score in each biopsy-matched urine sample within the violin plots and the distribution of scores is represented by the density shape of the plot. The thin black lines within each plot are the 25^th and 75^th percentile values, and the thick black line is the median value of the CTOT-04 signature score. The black horizontal dotted line crossing the contours of both violin plots is the CTOT-04 signature score of -1.213 which maximized sensitivity and specificity for distinguishing TCMR biopsies from NR biopsies in the CTOT-04 study.¹⁸

• Download figure
• Open in new tab

Figure 1. Violin Plots of the CTOT-04 Signature Scores in Urine matched to AR biopsies or NR biopsies.

Violin plots of CTOT-04 signature scores in the urine samples matched to AR biopsies (N=95) or urine samples matched to NR biopsies (N=50). Customized RT-qPCR assays were used to measure urinary cell levels of CD3E mRNA, CXCL10 mRNA, and 18s rRNA in kidney allograft biopsy-matched urine samples collected from kidney allograft recipients and the CTOT-04 signature score was calculated from log10-transformed values for the 18S-normalized CD3E mRNA, 18S-normalized CXCL10 mRNA and 18S rRNA.18 The CTOT-04 signature score values of individual urine samples are shown as circles within the plot. The two thin black lines within each plot are the 25th and 75th percentile values, and thick black line is the 50th (median) value of the CTOT-04 signature score. The horizontal dotted line parallel to the x-axis is the previously identified CTOT-04 signature score cutpoint of -1.213 that maximized sensitivity and specificity for the diagnosis of TCMR in the multicenter CTOT-04 study of 485 prospectively enrolled kidney allograft recipients.¹⁸

In the current study, the median (25^th percentile and 75^th percentile) CTOT-04 signature score was -0.480 (-1.450 and 0.570) in the 95 urine samples matched to 95 AR biopsies and significantly higher than the score of -1.945 (-2.668 and -1.200) in the 50 urine samples matched to 50 NR biopsies (Table 3; P<0.0001, T test). The previously validated cutpoint of -1.213 for TCMR biopsy diagnosis distinguished urine samples matched to AR biopsies consisting of TCMR, ABMR and MR from urine samples matched to NR biopsies at a specificity of 76% and a sensitivity of 71% (Table 4; P<0.0001, Fischer’s Exact test).

The median (25^th and 75^th percentiles) eGFR was 29 mL (18 and 38) in the AR group and 20 mL (11 and 33) in the NR group (P=0.0025). We examined whether the CTOT-04 signature score was associated with eGFR and determined that it was not associated (F-statistic: 0.3125, on 1 and 143 degrees of freedom, P=0.577).²³ We also found that the CTOT-04 signature score was diagnostic of AR after controlling for eGFR: P=3.88e-07 for the association without accounting for eGFR compared to P=6.4e-07 including eGFR in a general linear model.²⁴

The median time since transplantation to biopsy was 12.9 months (3.5 and 50.0) in the AR group and 2.4 months (0.4 and 6.0) in the NR group (P=0.004). We examined whether the CTOT-04 signature score was associated with time from transplantation to biopsy and found the score was marginally associated (F-statistic: 0.3503, on 1 and 143 degrees of freedom, P=0.06328).²³ We also evaluated whether the CTO-04 signature score was diagnostic of AR after controlling for time since transplantation to biopsy, and found it was diagnostic: P=3.88e-07 for the association without accounting for time since transplantation to biopsy compared to P=1.33e-06 including time in a general linear model.²⁴

CTOT-04 signature score discriminates kidney allograft recipients with NR biopsies from those with AR biopsies due to TCMR, ABMR or MR

The CTOT-04 signature, as expected from the earlier multicenter CTOT-04 study,¹⁸ was diagnostic of TCMR in the current investigation. Figure 2 shows the CTOT-04 signature score in each TCMR biopsy-matched urine sample. The median CTOT-04 signature score was -0.100 (-2.090 and 0.750) in the 47 urine samples matched to 47 TCMR biopsies and significantly higher than the score of -1.945 (-2.668 and -1.200) in the 50 urine samples matched to 50 NR biopsies (Table 3; P<0.0001, T test). At the CTOT-04 signature score cutpoint of -1.213, the signature distinguished urines matched to TCMR biopsies from urines matched to NR biopsies at a specificity of 76% and a sensitivity of 64% (Table 4; P<0.0001, Fischer’s Exact Test).

• Download figure
• Open in new tab

Figure 2. Violin Plots of CTOT-04 Signature Scores in Kidney Allograft Biopsy-Matched Urine Samples.

Violin plots of CTOT-04 signature scores in the urine samples matched to TCMR biopsies (n=47), ABMR biopsies (n=28), Mixed Rejection (MR) biopsies (n=20), or NR biopsies (N=50). Customized RT-qPCR assays were used to measure urinary cell levels of CD3E mRNA, CXCL10 mRNA, and 18s rRNA in kidney allograft biopsy-matched urine samples collected from kidney allograft recipients and the CTOT-04 signature score was calculated from log₁₀-transformed values for the 18S-normalized CD3E mRNA, 18S-normalized CXCL10 mRNA and 18S rRNA.¹⁸ The CTOT-04 signature score values of individual urine samples are shown as circles within the plot. The two thin black lines within each plot are the 25^th and 75^th percentile values, and thick black line is the 50^th (median) value of the CTOT-04 signature score. The horizontal dotted line parallel to the x-axis is the previously identified CTOT-04 signature score cutpoint of -1.213 that maximized sensitivity and specificity for the diagnosis of TCMR in the multicenter CTOT-04 study.¹⁸ One way analysis of variance comparing CTOT-04 signature score in urine samples matched to TCMR biopsies, ABMR biopsies, MR biopsies, or NR biopsies yielded a P value of <0.0001. Dunnett’s multiple comparison test for multiple group comparison yielded P <0.0001 for NR vs. TCMR; P <0.001 for NR vs. ABMR; and P <0.0001 for NR vs. MR.

The CTOT-04 signature score in each ABMR biopsy-matched urine sample is shown within the violin plot (Figure 2). The median CTOT-04 signature score in the 28 urine samples matched to 28 ABMR biopsies was -0.795 (-1.275 and 0.118) and significantly higher than the score of -1.945 (-2.668 and – 1.200) in the 50 urine samples matched to 50 NR biopsies (Table 3; P<0.001, T-test). At the CTOT-04 signature score cutpoint of -1.213, the signature distinguished urines matched to ABMR biopsies from urines matched to NR biopsies at a sensitivity of 75% and a specificity of 76% (Table 4; P<0.0001, Fischer’s Exact test).

The CTOT-04 signature score in each MR biopsy-matched urine sample is shown within the violin plot (Figure 2). The median CTOT-04 signature score in the 20 urines matched to 20 MR biopsies was 0.300 (-0.868 and 0.803) and significantly higher than the score of -1.945 (-2.668 and -1.200) in the 50 urines matched to 50 NR biopsies (Table 3, P<0.0001, T test). At the CTOT-04 signature score cutpoint of – 1.213, the signature distinguished urines matched to MR biopsies from urines matched to NR biopsies at a sensitivity of 80% and a specificity of 76% (Table 4; P<0.0001, Fischer’s Exact test).

The CTOT-04 signature score cutpoint of -1.213, developed for the diagnosis of TCMR,¹⁸ predicted TCMR, ABMR and MR in the current study with sensitivities and specificities that were not significantly different from the sensitivity and specificity for predicting TCMR in the CTOT-04 study (all comparisons, P>0.05).

Diagnostic performance of CTOT-04 signature assessed using AUCs of ROC curves

We determined the diagnostic performance of the CTOT-04 signature across different types of AR biopsies by comparing AUCs of ROC curves. For robust statistical inference, we performed bootstrap replications by drawing random samples with replacements from the original data. We generated 5000 bootstrap sample sets and estimated the AUCs of ROC curves. The CTOT-04 signature discriminated TCMR biopsies from NR biopsies with an optimism-corrected AUC of 0.749 (95% CI, 0.638 to 0.840; P<0.0001)(Figure 3A); distinguished ABMR biopsies from NR biopsies with an optimism-corrected AUC of 0.780 (95% CI, 0.656 to 0.878; P<0.0001) (Figure 3B); discriminated MR biopsies from NR biopsies with an optimism-corrected AUC of 0.857 (Bias-corrected 95% CI, 0.727 to 0.947; P<0.0001) (Figure 3C); and distinguished AR biopsies from NR biopsies with an optimism-corrected AUC of 0.780 (95% CI, 0.700 to 0.846; P<0.0001) (Figure 3D).

• Download figure
• Open in new tab

Figure 3. Receiver Operating Characteristic Curves For CTOT-04 Signature Scores in Kidney Allograft Biopsy-Matched Urine Samples.

Receiver operating characteristic curve CTOT-04 signature scores in the urine samples matched to TCMR biopsies (A), ABMR biopsies (B), MR biopsies (C), or AR biopsies (D). The fraction of true positive results (Sensitivity) and the fraction of false positive results (1-Specificity) for the CTOT-04 signature score calculated from log₁₀-transformed values for the 18S-normalized CD3E mRNA, 18S-normalized CXCL10 mRNA and 18S rRNA. A. In a comparison of patients with TCMR biopsies (n=47) with the patients with NR biopsies (n=50), the optimism-corrected AUC was 0.748 (Bias-corrected 95% CI, 0.638 to 0.840; P<0.0001). B. In a comparison of patients with ABMR biopsies (n=28) with the patients with NR biopsies (n=50), the optimism-corrected AUC was 0.780 (Bias-corrected 95% CI, 0.656 to 0.878; P<0.0001). C. In a comparison of patients with MR biopsies (n=20) with the patients with NR biopsies (n=50), the optimism-corrected AUC was 0.857 (Bias-corrected 95% CI, 0.727 to 0.947; P<0.0001). D. In a comparison of patients with AR biopsies (n=95) with the patients with NR biopsies (n=50), the optimism-corrected AUC was 0.780 (Bias-corrected 95% CI, 0.700 to 0.846; P<0.0001).

A comparison of the AUC of TCMR vs. NR to the AUC of ABMR vs. NR yielded a P value of 0.68; comparison of the AUC of TCMR vs. NR to the AUC of MR vs. NR yielded a P value of 0.15; and comparison of the AUC of TCMR vs. NR to the AUC of AR vs. NR yielded a P value of 0.61.

Altogether, our findings demonstrate that the CTOT-04 signature discriminates patients with NR biopsies from patients with AR, TCMR, ABMR or MR biopsies with similar ROC-AUC.