CV, IDV, and SRQ have a patent S14-048 (STAN-1110PRV) 61/951,908: Provisional Application pending. HAV, HL, CS, GC, and KKK have nothing to declare.
Conceived and designed the experiments: CV IDV SRQ HAV KKK. Performed the experiments: CV IDV HL CS GC LP. Analyzed the data: CV IDV. Wrote the first draft of the manuscript: CV IDV SRQ KKK. Contributed to the writing of the manuscript: CV IDV SRQ KKK HAV HL CS LP GC. Enrolled patients: HAV HL CS GC KKK. Agree with the manuscript’s results and conclusions: CV IDV SRQ KKK HAV HL CS LP GC. All authors have read, and confirm that they meet, ICMJE criteria for authorship.
It remains difficult to predict and to measure the efficacy of pharmacological immunosuppression. We hypothesized that measuring the B-cell repertoire would enable assessment of the overall level of immunosuppression after heart transplantation.
In this proof-of-concept study, we implemented a molecular-barcode-based immune repertoire sequencing assay that sensitively and accurately measures the isotype and clonal composition of the circulating B cell repertoire. We used this assay to measure the temporal response of the B cell repertoire to immunosuppression after heart transplantation. We selected a subset of 12 participants from a larger prospective cohort study (
If confirmed in larger, prospective studies, the method described here has potential applications in the tailored management of post-transplant immunosuppression and, more broadly, as a method for assessing the overall activity of the immune system.
A novel, proof-of-concept, diagnostic test for allograft rejection. The authors use a barcode-sequencing approach to determine the isotypes and clonal composition of circulating B cells in heart transplant patients.
Fifty years ago, little could be done to help a person whose heart, liver, or kidneys had stopped working because of infection, injury, or a chronic disease process. Nowadays, solid organ transplantation gives more than 100,000 people worldwide a new lease on life every year. Many different organs can be successfully moved from one individual (the organ donor), who may be dead or alive, into another individual (the organ recipient). However, organ transplantation remains a complex and risky procedure. The organ recipient may not survive transplantation surgery, and sometimes the transplanted organ fails to work in its new owner. Another major risk for patients who receive an allograft (an organ from another person) is acute rejection. Our immune system protects us from pathogens (disease-causing organisms) by recognizing specific molecules (antigens) on an invader’s surface as foreign and initiating a sequence of events that kills the invader. Unfortunately, the immune system sometimes recognizes organ transplants as foreign and triggers acute transplant rejection. The chances of rejection can be minimized by “matching” the antigens on the donated organ to those on recipient’s organs and by giving the recipient immunosuppressive drugs such as prednisone and tacrolimus.
The primary aim of post-transplantation immunosuppression is to prevent acute rejection, which is associated with the activation of T cells (a type of lymphocyte or white blood cell). But immunosuppressive therapy also affects other immune system cells, including B cells, another type of lymphocyte. B cells make antibodies, molecules that recognize foreign antigens and tag pathogens for destruction by the immune system. It is important, therefore, to get the level of immunosuppression just right for every patient—too little and the transplanted organ will be rejected, too much and the patient will be susceptible to infections. In this proof-of-concept study, the researchers ask whether measurement of the B cell repertoire (the mixture of antibodies produced by a person’s B cells) can be used to assess the overall level of immunosuppression in patients after heart transplantation. When the body is threatened by a pathogen, naïve B cells undergo molecular processes called hypermutation and class-switch recombination. Hypermutation allows the immune system to adapt to new threats by changing the specificity and strength of antibody binding to antigens and involves changes in the “variable” regions of antibodies. Class-switch recombination changes the “isotype” of the antibody produced by a B cell by switching the constant region of the antibody’s “heavy chain.” Thus, determination of the composition of immunoglobulin heavy chain transcripts (nucleotide sequences that are translated into antibodies) in the blood could indicate the level of immunosuppression in an individual.
The researchers obtained blood samples collected over several months from twelve patients who had had a heart transplant, including six who had experienced a rejection event. Using a previously developed barcode-based immune repertoire sequencing assay, the researchers measured changes over time in the sequence and abundance of immunoglobulin heavy chains in the patients. They found that the fraction of antibody sequences related to activated B cells was inversely related to the level of tacrolimus in the patients’ blood and that antibody repertoire sequencing could detect immune activation during acute rejection events diagnosed by examining small pieces of the donated heart or by looking for donor-derived DNA in the recipients’ blood. Specifically, the immune repertoire sequencing assay diagnosed acute allograft rejection with a sensitivity of 71.4% and a specificity of 82.0%. That is, the assay correctly identified 71.4% of patients who had had a rejection event and 82.0% patients who had not had a rejection event.
These findings suggest that measurement of the immune repertoire using a high-throughput sequencing assay might provide an accurate and sensitive way to manage immunosuppression in patients who have received an organ or tissue transplant. Importantly, however, these findings provide only proof of concept. Large prospective studies are now needed to confirm the ability of immune repertoire sequencing to predict rejection events and to compare this approach to existing assays for transplant rejection. If validated, immune repertoire sequencing may eventually make it possible to personalize immunosuppressive therapy after transplantation to ensure adequate suppression of the immune system’s response to the allograft while avoiding the many long-term consequences of over-immunosuppression.
This list of resources contains links that can be accessed when viewing the PDF on a device or via the online version of the article at
The UK National Health Service provides information about
The British Heart Foundation, a not-for-profit organization, also provides information about
The US National Heart, Lung, and Blood Institute provides further information about
The British Liver Trust, a not-for-profit organization, explains why
Wikipedia has pages about
The MedlinePlus Medical Encyclopedia has pages on
An enduring challenge in immunology is the lack of quantitative measurements of immune strength. Current clinical practice relies on very crude estimates of the activity of the immune system, such as white blood cell counts. In view of the lack of more predictive assays, pharmacological immunosuppressive therapy, e.g., in the context of post-organ-transplant therapy, is guided mainly by dosage and measurement of the concentration of immunosuppressive drugs in blood. In adult transplant recipients, these immunosuppressive drugs typically include induction agents (e.g., lymphocyte-depleting antibodies, such as anti-thymocyte globulin) followed by maintenance with a combination of corticosteroids (prednisone), calcineurin inhibitors (tacrolimus and cyclosporine), and anti-proliferative agents (most commonly mycophenolate mofetil [MMF]). Immunosuppression is therefore achieved by combining several drugs with distinct mechanisms of action. Calcineurin inhibitors, for example, inhibit or deplete T helper and T killer cells, respectively, and consequently reduce T-helper-cell-dependent B cell activation [
Measuring the activity of the immune system directly would allow a more comprehensive understanding of the net state of the immune system. Here, we tested the hypothesis that immune repertoire sequencing of the B cell antibody heavy chain could provide an accurate and individualized measure of the activity of the adaptive immune response. The antibody heavy chain IGH transcript is unique as its expression changes fundamentally, not only in abundance but also in sequence, when a B cell is activated. B cells can undergo hypermutation and class-switch recombination when activated. Activated B cells express high levels of mutated antibodies of the IgG and IgA isotypes, while naïve B cells express non-mutated IgM antibodies at low levels (
Organ transplant recipients receive immunosuppressive therapy, which usually includes calcineurin inhibitors. These drugs inhibit T cell activation, leading to fewer B cells expressing class-switched and mutated IGH transcripts. We measured overall immunosuppression using immune repertoire sequencing (IGH-Seq). A schematic of the IGH-Seq protocol and data analysis is shown on the right. C_Region, primer pool specific to constant region; CDR3, complementary determining region 3; PBMC, peripheral blood mononuclear cell; UID, unique identifier; V_FR3, primer pool specific to V segments.
Immune repertoire sequencing relies on high-throughput sequencers to identify the composition of IGH transcripts in a blood sample, and has been used for various applications, including tracking of minimal residual disease in acute lymphoblastic leukemia, studying age-related changes in the immune system, and identifying the memory B cell response to influenza vaccination [
Here, we determined whether the immune repertoire sequencing approach enables quantification of loss of immune activity in response to pharmacological immunosuppression. We found that the fraction of antibody sequences related to activated B cells is inversely correlated with the dosage of immunosuppressive drugs, and could be used as a marker of net immunosuppression. We furthermore found that antibody repertoire sequencing detects immune activation during acute rejection events, as diagnosed by cell-free donor-derived DNA (cfdDNA) levels or endomyocardial biopsies [
This work was approved by the Stanford University Institutional Review Board (protocol 17666), and all individuals provided written informed consent.
The specific goal of this proof-of-concept study was to test the clinical utility of immune repertoire sequencing to measure immunosuppression and predict acute rejection after heart transplantation. Participants included in this study were selected from a larger, ongoing prospective cohort study that involves recruitment of de novo pediatric and adult heart transplant candidates at Stanford Medical Center. The original cohort study was designed to determine the utility of cfdDNA for acute rejection surveillance after heart transplantation, and has been described previously [
We selected 12 adult transplant recipients from this larger prospective cohort study to represent patients with and without acute rejection events who were otherwise free of severe infections and other immune-related disorders. All buffy coat samples available at the onset of the study were analyzed, resulting in an average of 15 mo of follow-up for the 12 participants. Samples collected after this point were not included in the study. The main outcome measures of this study were the abundance and composition of the circulating B cell repertoire, in order to determine the temporal response of the B cell repertoire to immunosuppression after heart transplantation. We were interested in investigating the temporal dynamics of the B cell repertoire and therefore selected patients for whom a relatively large number of longitudinal samples were available. Among the patients analyzed in this study, six had experienced a rejection-free post-transplant course (patients 1–6; endomyocardial biopsy grade < 2R and cfdDNA below a previously determined threshold of 1% donor DNA), and six patients had experienced a moderate-to-severe rejection event as determined by endomyocardial biopsy grade ≥ 2R or elevated cfdDNA levels (patients 7–12) (
Post-transplant immunosuppression consisted of methylprednisolone 500 mg administered immediately postoperatively, followed by 125 mg every 8 h for three doses. Anti-thymocyte globulin (rATG) 1 mg/kg was administered on postoperative days 1, 2, and 3. Maintenance immunosuppression consisted of prednisone 20 mg twice daily starting on postoperative day 1 and tapered to <0.1 mg/kg/d by the sixth postoperative month, and tapered further if endomyocardial biopsies showed no evidence of cellular rejection. Tacrolimus was started on postoperative day 1, and dosing was adjusted to maintain a level of 10–15 ng/ml during months 0–6, 7–10 ng/ml during months 6–12, and 5–10 ng/ml thereafter. MMF was started at 1,000 mg twice daily on postoperative day 1, and dose adjustments were made, if required, in response to leukopenia. All heart transplant recipients, except for those in whom both the donor and recipient had no history of prior cytomegalovirus infection (i.e., donor and recipient cytomegalovirus IgG-negative), received anti-viral prophylaxis with valganciclovir 900 mg twice daily for 2 wk, then 900 mg daily until 6 mo post-transplant, followed by 450 mg daily until 12 mo post-transplant, at which point anti-viral prophylaxis was discontinued. Valganciclovir dose reductions were made in the setting of leukopenia. Anti-fungal prophylaxis consisted of itraconazole 300 mg daily for the first 3 mo post-transplant, and prophylaxis against
Plasma and buffy coat samples were extracted from whole blood by sequential centrifugation within 3 h of sample collection and stored at −80°C. When required for analysis, plasma samples were thawed, and circulating DNA was immediately extracted from 0.5–1 ml of plasma using the QIAamp Circulating Nucleic Acid Kit (Qiagen). RNA was extracted from buffy coat samples using TRIzol Reagent (Life Technologies) followed by RNeasy column (Qiagen) extraction.
Libraries were generated by a single researcher who was blinded to all clinical data until after sequencing and initial data analyses had occurred. Sequencing libraries were prepared from the purified patient plasma DNA using the NEBNext DNA Library Prep Master Mix Set for Illumina with standard Illumina indexed adapters, or using a microfluidics-based automated library preparation platform (Mondrian SP Ovation SP Ultralow Library System, Nugen). Libraries were characterized using the Agilent 2100 Bioanalyzer (high-sensitivity DNA kit) and quantified by quantitative PCR.
Libraries were generated by two investigators (L. P. and C. V.) who were blinded to all clinical data until after sequencing and initial data analyses had occurred. 0.5–1 μm of total RNA were used as input for library preparation. Reverse transcription used SuperScript III Enzyme (Life Technologies) and primers for all five isotype constant regions containing eight random nucleotides and partial Illumina adapters. Second-strand synthesis was done using Phusion High-Fidelity DNA Polymerase (New England Biolabs) and primers for framing region 3 off all IGH variable (V) segments containing eight random nucleotides and partial Illumina adapters. The combined 16 random nucleotides (eight constant region primers and eight V segment primers) serve as a unique identifier (UID) for each IGH transcript molecule. Double-stranded cDNA was purified two times using AMPure XP beads at a ratio of 1:1. Double-stranded cDNA was amplified with Platinum Taq DNA Polymerase High Fidelity (Life Technologies) and two custom Illumina indexing primers completing the Illumina adapters and containing two Illumina index sequences (i5 and i7). Final sequencing libraries were generated by purifying the PCR product using AMPure XP beads at a ratio of 1:1 (
Immune repertoire libraries were grouped into library pools with up to 50 libraries with compatible Illumina indices and sequenced on the Illumina HiSeq 2000 or MiSeq using 2 × 100 bp or 2 × 150 bp protocols. The sequencing runs yielded between 1 and 10 million raw reads per sample. Raw reads were split into UID groups with unique 16-nucleotide identifiers. Every UID group defined a single original IGH transcript molecule, and consensus reads were built for forward and reverse reads of each UID group, correcting sequencing errors for UID groups with coverage of two or more raw sequencing reads. While this approach does not result in the correction of sequencing errors for UID groups with a coverage of one, it still identifies these groups as originating from a unique molecule, thereby aiding in quantification. Forward consensus reads were annotated by alignments to V segments using BLAST, yielding mutation rate information. Reverse consensus reads were aligned to constant regions using BLAST [
Shotgun DNA sequencing reads generated from the cfdDNA libraries that did not map to the human genome were aligned to a reference of viral, bacterial, and fungal genomes (BLAST). The relative genomic abundance of Anelloviridae virus was calculated with GRAMMy, a tool that utilizes BLAST-derived nucleic acid sequence similarity data to estimate the relative abundance of species [
Differences between groups (no evidence of rejection, mild acute rejection diagnosed by endomyocardial biopsy, moderate-to-severe acute rejection diagnosed by endomyocardial biopsy, and rejection diagnosed by elevated cfdDNA) were tested using Mann-Whitney U tests. C-statistic, sensitivity, and specificity were determined using the sklearn.metrics package for Python [
(A) Relative IGH sequence composition at increasing expression levels (molecules/sequence) is shown across all individuals (
The shotgun DNA sequence data have been deposited in the Sequence Read Archive (SRP034946). The immune repertoire sequence data have been deposited in the Sequence Read Archive (PRJNA260905).
In this proof-of-concept study, we analyzed 130 blood samples from 12 patients who underwent pharmacological immunosuppression to prevent allograft rejection after heart transplantation. These individuals were a subset of a larger prospective cohort study [
We first determined whether we could measure the effects of immunosuppression on the activity of the immune system using immune repertoire sequencing of the antibody IGH transcript. To this end, we pooled data for all participants in the study and separated antibody IGH transcript sequences by expression level.
Lowly expressed sequences were defined as antibody IGH transcript sequences that were represented by only one molecule and that were likely expressed by naïve or inactive B cells. Highly expressed sequences were defined as antibody IGH transcript sequences that were represented by more than one molecule and that were likely expressed by activated B cells such as plasmablasts [
We compared all data, including time points with diagnosed rejection events, pooled from the 12 participants of the main study at the beginning of (day 1) and 6 mo into (day 180, 20 d before first rejection event) immunosuppressive therapy. When normalized to the total number of molecules analyzed, the number of highly expressed sequences did not vary substantially between day 1 (2,131 highly expressed sequences/19,845 molecules; ratio 0.107) and day 180 (8,018 highly expressed sequences/79,852 molecules; ratio 0.100). Interestingly, a much larger change was observed in the composition of highly expressed sequences. Highly expressed IgA, IgD, IgE, IgG, and mutated IgM sequences decreased from 1,296/19,845 molecules (ratio 0.065) at day 1 to 847/79,852 molecules (ratio 0.011) by day 180 after the induction of immunosuppression (
We found that ABS level decreased rapidly in the first month, and the initial rapid decrease was followed by a slow decline over subsequent months. ABS level approached approximately 15% of the original level after 10 mo, after which the ABS level increased in response to tapering of immunosuppressive therapy. The decrease in ABS level was strongly and negatively correlated with blood tacrolimus concentration (Pearson
By reducing the activity of the immune system, immunosuppressive therapy also alters selective pressure on bacterial and viral populations in the human body. We previously showed that the composition of the viral population, as measured by sequencing of cell-free DNA in plasma, is significantly affected by post-transplant immunosuppressive therapy [
Next, we examined whether ABS levels are also informative of acute rejection events. We examined ABS level in four patients, two of whom experienced acute rejection as diagnosed by either cfdDNA level (>1% cfdDNA) or endomyocardial biopsy. While the ABS level of individuals without rejection remained at baseline throughout the study period (
ABS levels are shown for two individuals with (C and D) and two individuals without (A and B) moderate-to-severe acute rejection events (top). Colors indicate isotype contribution. cfdDNA levels are shown for the same individuals (bottom). Biopsy grades are shown as vertical blue lines (bottom). Thin lines indicate mild ACR events (grade 1R), and thick lines indicate moderate-to-severe ACR events (grade ≥ 2R).
To investigate the performance of the immune repertoire sequencing assay compared to cfdDNA and biopsy measurements, we plotted the ABS levels of all 12 study patients, highlighting acute rejection events as diagnosed by cfdDNA and endomyocardial biopsy (Figs
(A) ABS levels plotted for all study participants. Acute rejection events as diagnosed by endomyocardial biopsy (ACR, violet; acute antibody-mediated rejection [AMR], green) or cfdDNA level (pink) are shown in color. (B) Samples are classified by rejection status diagnosed by biopsy as no evidence of rejection, mild ACR (biopsy grade 1R), or moderate-to-severe ACR (biopsy grade ≥ 2R), and by rejection status diagnosed by the cfdDNA assay. Samples are colored as in (A) if at any point the organ recipient experienced a serious rejection event. As biopsy grades are independent of cfdDNA measurements, there is overlap between samples in the elevated cfdDNA and biopsy-proven rejection bins. The groups were compared using a one-sided Mann-Whitney U test, which has high efficiency for both normally and non-normally distributed datasets. (C) Receiver operating characteristic (ROC) curves for ABS levels compared to cfdDNA (pink) and endomyocardial biopsy (violet) results. FPR, false rejection rate; TPR, true positive rate. (D) ABS levels measured before and after rejection diagnosis. The black line shows the single-exponent fit of ABS levels measured prior to rejection,
For a systematic analysis, we grouped samples according to biopsy and cfdDNA results. While all endomyocardial biopsies were tested for ACR, only some biopsies were also tested for antibody-mediated rejection, based on clinical parameters, thereby preventing systematic analysis for this type of acute rejection. In the cohort studied, antibody-mediated rejection was clinically reported at only one time point (
To investigate the potential for early diagnosis, we analyzed the temporal dynamics of ABS level prior to a biopsy-proven moderate-to-severe rejection event (
We next sought to examine the effect of opportunistic infections and B cell disorders on antibody repertoire sequencing measurements. To this end, we tracked ABS level in 25 samples from two additional heart transplant recipients. The first patient experienced several infections diagnosed both before and after transplantation, including cytomegalovirus, Epstein-Barr virus, hepatitis B virus, varicella zoster virus, staphylococcus, parainfluenza 3, and
ABS level is shown for three individuals: (A) a heart transplant recipient without moderate-to-severe rejection events, (B) a heart transplant recipient with infectious complications, and (C) a heart transplant recipient with AL amyloidosis. For the patient with AL amyloidosis (C), the approximate schedule of drug treatment is indicated at the top of the panel. Colors indicate isotype contribution. cfdDNA level is shown for the same individuals (bottom). Biopsy grades are shown as vertical blue lines (bottom): thin lines indicate mild rejection events (grade 1R), and thick lines indicate moderate-to-severe rejection events (grade ≥ 2R). SCT, stem cell transplant.
Heart transplantation is considered the best therapeutic option for the treatment of eligible patients with end-stage heart disease [
Samples were collected longitudinally from the heart transplant recipients in this study, and immune repertoire measurements were compared to biopsy grades, tacrolimus levels, and the level of cfdDNA, as determined by shotgun sequencing. This longitudinal sampling enabled us to closely monitor the dynamics of immunosuppression. Measurements of cfdDNA and biopsies performed in parallel to the immune repertoire assay made it possible to compare our assay to both a noninvasive independent assay (cfdDNA) as well as the invasive diagnostic gold standard for acute organ rejection (endomyocardial biopsy). This approach provided us with an unprecedented view of the immune response to pharmacological immunosuppression, and how it escapes sufficient suppression, resulting in acute rejection.
Several commercially available assays aim to accomplish these goals [
The immune repertoire assay presented here is very efficient in terms of both sample preparation and sequencing: in this study we implemented a dual indexing strategy that allowed us to multiplex more than 200 samples on a single HiSeq 2000 flow-cell lane, thereby reducing the cost per sample—including sample preparation—to approximately US$40. The assay furthermore provides a value based on internal controls that does not rely on external normalization, and can be performed on small and low-quality samples. In addition to our main focus on immunosuppression and immune activation via rejection, we examined the dynamics of the B cell repertoire for two patients with unusual post-transplant courses: one patient with severe infectious complications and one patient with a B cell disorder. These case studies indicate the potential of immune repertoire sequencing in the management of a diverse set of post-transplant complications and also illustrate the potential value of combining immune repertoire sequencing with other assays, such as the measurement of cfdDNA in plasma, in order to discriminate between possible causes of immune activation.
Prospective studies are required to validate this assay in a larger patient population, to compare immune repertoire sequencing assay results to other noninvasive screening tests for heart transplant complications (cfdDNA, AlloMap), and to investigate the utility of this assay in the setting of other solid organ transplants (e.g., lung, kidney, and liver transplantation). Finally, studies that use immune repertoire sequencing results to tailor immunosuppression in individual patients may enable a “personalized medicine” approach to achieving adequate suppression of the alloimmune response, while avoiding the myriad long-term sequelae of over-immunosuppression.
(GZ)
ABS levels (top) are shown for six individuals with and six individuals without moderate-to-severe acute rejection events, as well as for one individual with chronic infection and one individual with amyloidosis. Colors indicate isotype contribution. cfdDNA levels are shown for the same individuals (bottom). Biopsy grades are shown as vertical blue lines (bottom): thin lines indicate mild ACR events (grade 1R), and thick lines indicate moderate-to-severe ACR events (grade ≥ 2R).
(TIFF)
IGH molecules are grouped by V and J segment usage and shown as a scatter plot. Every dot represents a V–J combination, with its size indicating the number of molecules featuring that particular V–J combination. Plots are shown for three different time points (onset of post-transplant immunosuppressive therapy [day 1], maximal immunosuppression [day 104], and expansion of IgA [Day 584]). Even distribution of V–J usage suggests polyclonal expansion of IgA sequences.
(TIF)
(PDF)
(PDF)
We thank G. Mantalas, S. Liu, J. Okamoto, and N. Neff for assistance with the sequencing assays and B. Passarelli for setting up and maintaining the computing infrastructure.
activated B cell sequence
acute cellular rejection
amyloid light chain
cell-free donor-derived DNA
mycophenolate mofetil
receiver operating characteristic
unique identifier
variable