As HLAs antibody detection technology has evolved, there is now detailed HLA antibody information available on prospective transplant recipients. Determining single antigen antibody specificity allows for a calculated panel reactive antibodies (cPRA) value, providing an estimate of the effective donor pool. For broadly sensitized lung transplant candidates (cPRA ≥ 80%), our center adopted a pretransplant multi-modal desensitization protocol in an effort to decrease the cPRA and expand the donor pool. This desensitization protocol included plasmapheresis, solumedrol, bortezomib and rituximab given in combination over 19 days followed by intravenous immunoglobulin. Eight of 18 candidates completed therapy with the primary reasons for early discontinuation being transplant (by avoiding unacceptable antigens) or thrombocytopenia. In a mixed-model analysis, there were no significant changes in PRA or cPRA changes over time with the protocol. A sub-analysis of the median fluorescence intensity (MFI) change indicated a small decline that was significant in antibodies with MFI 5000–10 000. Nine of 18 candidates subsequently had a transplant. Posttransplant survival in these nine recipients was comparable to other pretransplant-sensitized recipients who did not receive therapy. In summary, an aggressive multi-modal desensitization protocol does not significantly reduce pretransplant HLA antibodies in a broadly sensitized lung transplant candidate cohort.
calculated panel reactive antibody
median fluorescence intensity
panel reactive antibody
Over the last decade, the number of lung transplant procedures has steadily increased as centers are considering older candidates, ventilator-dependent candidates and extracorporeal membrane oxygenation candidates [1-4]. However, a key factor limiting this growth is the presence of HLA antibodies in transplant candidates. These HLA antibodies specific to HLA class I antigens, which are expressed on most nucleated cells, or to HLA class II antigens, which are expressed on antigen-presenting cells, can initiate a cascade of complement and neutrophil directed inflammation at the time of transplant if the donor organ expresses the specific HLA . In order to avoid antibody-mediated damage to the endothelium, interstitial edema and fibrin thrombi formation in the allograft [6, 7], it is now common practice in lung transplantation to consider a donor as unacceptable if the transplant candidate has antibodies specific to that donor HLA.
HLA antibody detection has evolved from complement-dependent lymphocytotoxicity (CDC)-based tests to solid-phase immunoassays with single antigen bead arrays for antibody detection . The fine specificity of the currently available single antigen antibody test allows for a calculated panel reactive antibodies (cPRA) value, providing an estimate of the effective donor pool. In lung transplant candidates with a high cPRA and thus a small available donor pool, the waiting time and risk of death on the transplant waitlist is considerable. Thus, centers have employed several therapeutic approaches in an effort to lower or “desensitize” HLA antibody positive individuals prior to transplant.
Several methods and medications can be used to remove antibodies (Table 1). In renal transplant, which has the highest rates of antibody-sensitized candidates, desensitization protocols that include a combination of modalities may yield better outcomes, particularly with rejection after transplant. The two most common combinations are plasmapheresis with intravenous immunoglobulin (IVIG) [9-11] or rituximab with IVIG [12-14]. Results have been mixed with some reports of antibody reduction and successful transplant with negative cross-match and other reports with no change in antibodies. Of note, protocols have been widely variable in the number and timing of treatments as well as the number of candidates treated.
|Desensitization strategy||Therapeutic class||Approximate cost1||Mechanism of action||Therapeutic effect|
|Alemtuzumab||Monoclonal Ab||$653 for 10 mg, though currently $0 for transplant||Binds to CD52 on mature lymphocytes||Depletes circulating T and B cells|
|Bortezomib||Chemotherapy||$44.89 for 0.1 mg||Proteasome inhibitor||Promotes apoptosis in plasma cells|
|Cyclophosphamide||Chemotherapy||$39.67 for 100 mg||Alkylating agent||Myelosuppressive agent|
|Eculizumab||Monoclonal Ab||$201.12 for 10 mg||Complement (C5) inhibitor||Inhibits complement-mediated toxicity|
|Intravenous immunoglobulin||Pooled polyclonal Abs||$36–$40 for 500 mg||Multiple (see text)||Generalized immunosuppression|
|Plasma exchange||Procedure||Removal of circulating Abs||Generalized immunosuppression|
|Rituximab||Monoclonal Ab||$678.70 for 100 mg||Binds to CD20 on B cells||Depletes circulating B cells|
|Splenectomy||Procedure||Removal of secondary lymphoid organ||Generalized immunosuppression|
In thoracic transplant, there are limited data using multi-modal antibody desensitization. In a case series of six heart transplant candidates, rituximab/IVIG combination therapy failed to lower HLA antibodies levels, but after the addition of bortezomib and plasmapheresis, the cPRA was lower and four of the six candidates were subsequently transplanted with negative cross-matches. Of note, a positive antibody was defined as a median fluorescence intensity (MFI) >5000 for this study and while the cPRA changed, the breadth of the sensitization did not necessarily change . A report of two lung transplant candidates utilizing plasmapheresis, IVIG, rituximab and bortezomib suggested that combination therapy may be useful in decreasing antibodies temporarily; however, HLA antibodies subsequently rebounded prior to transplant .
In another aggressive antibody reduction protocol, renal recipients with antibody-mediated rejection were given plasmapheresis, bortezomib and rituximab in multiple doses over a period of 18 days . Shortly after this publication, our center instituted a similar protocol for desensitization in broadly sensitized lung transplant candidates. In this retrospective analysis, we examine the efficacy of using a multi-modal therapy prior to lung transplantation in candidates with combined class I and class II HLA that corresponded to cPRA ≥ 80%. In addition, we reviewed tolerability and complications that led to early discontinuation of therapy. Finally, we evaluated 1-year mortality in those candidates who subsequently had a lung transplant.
Materials and Methods
Lung transplant candidates with combined HLA class I and class II antibodies corresponding to cPRA ≥80% were eligible for the multi-modal desensitization protocol. Individuals were followed in an intention to treat analysis. The antibody analysis period spanned the time of first HLA test at our center until transplant, last HLA antibody test if lost to follow-up or May 1, 2013, whichever occurred first. For the posttransplant survival analysis, all lung transplant recipients transplanted after January 1, 2010 and before the censor date of May 1, 2013 at Duke University Medical Center with detectable HLA antibodies pretransplant were included. Retransplant, multi-organ, living lobar and pediatric recipients were excluded in the survival analysis. The posttransplant survival period was from transplant until date of death or censor. Censoring events for the survival analysis were retransplant, transfer of care to another transplant center or May 1, 2013, whichever occurred first. This retrospective analysis was approved through the Duke University IRB (IRB# 00044155).
Clinical protocol posttransplant
All posttransplant recipients with detectable pretransplant HLA antibodies are given weekly IVIG (500 mg/kg) × 6 weeks, then monthly IVIG × 3 months followed by IVIG every 3 months. Induction immunosuppression includes basiliximab and methylprednisone. Maintenance immunosuppression includes tacrolimus, prednisone and mycophenolate mofetil.
HLA antibody measurements
Our center protocol calls for HLA antibody tests pretransplant at all clinic visits (typically every 1–3 months) and then monthly while on the transplant waiting list. There are intermittent single antibody specificities performed for cPRA determinations. For percentage of class I and class II HLA, panel reactive antibody (PRA) beads coated with purified HLA class I or class II antigens were incubated with the patient's serum according to the manufacturer's protocol, stained with fluorescein isothiocyanate-conjugated with anti-human IgG and analyzed using a flow cytometer (One Lambda, Inc., Canoga Park, CA). Subsequent testing with single antigen beads (Luminex; One Lambda) provides fine antigen specificities (97 class I antigens and 91 class II antigens). Our center uses MFI = 1000 as a standard cutoff for a positive antibody binding. The established positive MFI value for Luminex single antigen bead assay was based on comparisons with flow cytometric single antigen bead assay, correlations with flow cross-match and the recommendation of the reagent vendor. Our center requires a repeat sample to confirm all de novo HLA antibodies.
Plasmapheresis was performed using a COBE® Spectra or Spectra Optia System (Terumo BCT, Lakewood, CO). A 2000–3000 mL exchange was performed with 5% albumin replacement prior to administration of chemotherapy. Other components of the multi-modal desensitization protocol included methylprednisolone, bortezomib, rituximab and IVIG as shown in Figure 1. Candidates included in the analysis received all modalities, but not all patients completed the standard protocol, due to patient factors and protocol-specific events (see “Tolerability and Complications”). Patients were screened for adverse events including thrombocytopenia, neuropathy and infection. Therapy plans were adjusted accordingly in the setting of an adverse event. Bortezomib was initially given intravenously, but changed to subcutaneous administration due to decreased risk of neuropathy.
Cohort demographics were summarized with descriptive statistics. Repeated measurements analyses for PRA were used with mixed models to determine if multi-modal therapy reduced class I and class II antibodies. class I and class II percentages were considered in separate models. A spatial data covariance (error) structure allowed for irregular time intervals between measurements and differing number of measurements per subject. Time was set relative to the initiation of desensitization protocol. The therapy effect term was categorized as “complete” if at least six cycles of plasmaspheresis, four doses of bortezomib and one dose of rituximab were administered and “incomplete” if fewer were given or “none” if prior to therapy start. The effect of therapy on a HLA test was considered with time from therapy start. A secondary mixed-model analysis of HLA antibody changes by MFI was done to consider high MFI (>10 000), moderate MFI (5000–10 000) and low MFI (<5000) antibody changes for both class I and class II antibodies. In this analysis, MFI <1000 were included. Survival after transplant was compared between presensitized candidates who underwent transplant without desensitization therapy and those who underwent the above desensitization protocol in an unadjusted Kaplan–Meier analysis.
Primary endpoints were changes in HLA class I antibodies and HLA class II antibodies over time considering both PRA and cPRA prior to transplant. Secondary end points were change in HLA antibodies by MFI, 1-year survival posttransplant in those who proceeded to transplant, protocol tolerability and protocol complications. Analysis was completed in SAS 9.2 (Cary, NC).
Eighteen lung transplant candidates were initiated on the desensitization protocol. Nine of 18 these candidates subsequently went to transplant with three of these having an unexpected positive retrospective cross-match thought to be due to medication interference and/or non-HLA antibodies. Median wait time for these candidates was 93 days (interquartile range [IQR] 81, 267) compared with our center median wait time of approximately 12 days. Two of 18 candidates are currently listed for transplant at our center. The remaining seven candidates are no longer being considered for transplant at our center due to a variety of medical reasons. In the posttransplant survival analysis, there were 114 lung transplant recipients with pretransplant detectable HLA antibodies who did not undergo the desensitization protocol and constituted the comparison group; this group was chosen as the comparison group given the association of pretransplant HLA antibodies with worse survival posttransplant. A description of both groups is summarized in Table 2.
|Pretransplant sensitized (n = 114)||Pretransplant-sensitized and multi-modal therapy (n = 18)||p-Value|
|Gender, female, n (%)||51 (89)||13 (72)||0.20|
|Race, n (%)|
|White||107 (94)||14 (78)||0.70|
|Black||4 (4)||3 (17)|
|Other||3 (2)||5 (5)|
|Native lung disease, n (%)|
|Cystic||11 (10)||1 (11)||0.62|
|Obstructive||29 (26)||4 (44)|
|Restrictive||70 (62)||4 (44)|
|Vascular||3 (2)||0 (0)|
|HLA percentages pretransplant, median (IQR)|
|Class I PRA||8 (5, 15)||75 (19, 99)||<0.0001|
|Class I cPRA||12 (4, 34)||92 (73, 98)||<0.0001|
|Class II PRA||0 (0, 9)||76 (9, 87)||<0.0001|
|Class II cPRA||29 (9, 59)||85 (74, 97)||<0.0001|
|HLA percentages pretransplant, mean ± SD|
|Class I PRA||13.2 ± 16.3||63 ± 38.4||<0.0001|
|Class I cPRA||20.6 ± 20.5||81 ± 21.9||<0.0001|
|Class II PRA||9.7 ± 19.3||56.4 ± 39.5||<0.0001|
|Subsequent transplant, n (%)||114 (100)||9 (50)|
|Days from therapy to transplant, median days (IQR)||n/a||69 (38, 129)|
|Age at transplant, median years ± standard deviation||64 ± 14.6||58.0 ± 12.5||0.51|
|Received complete therapy, n (%)||n/a||8 (44)|
|Posttransplant follow-up days, median days (IQR)||579 (346, 798)||386 (249, 775)||0.74|
HLA antibodies by class
There were 197 pretransplant HLA antibody tests available for analysis in the HLA positive candidates undergoing desensitization cohort. Median number of tests per recipient was 11 (IQR 6, 16). Median PRA percentage class I was 59% (IQR 11%, 86%) and class II was 67% (18%, 87%). Median cPRA percentage class I was 73% (IQR 54%, 97%) and class II was 69% (61%, 91%). The majority (16 of the 18 candidates) had both class I and class II antibodies detected. One candidate had just class I and one candidate had just class II antibodies.
Class I percentage pretransplant HLA antibodies did not significantly change over time with the complete desensitization intervention, though there was a minimal change for incomplete therapy (p = 0.05 if partial therapy and p = 0.21 if complete therapy). Class II percentage antibodies did not show a statistically significant change over time with partial or complete therapy (p = 0.6 and 0.64, respectively). Individual HLA antibody measurements over time are included in Figure A1 in the Appendix.
Single antigen determination with cPRA was available for 122 cPRA class I and 113 cPRA class II values. Median cPRA class I was 73% (IQR 54%, 97%) and class II was 69% (IQR 61%, 91%). Class I cPRA percentages increased slightly in individuals with partial therapy (estimate 0.11% per day) but had no change for complete therapy (p = 0.28). Class II cPRA did not change over time with therapy for both partial therapy (p = 0.64) and complete therapy (p = 0.66).
HLA antibody changes by MFI
For each candidate, a high, moderate and low MFI antibody was selected from the class I and class II detected antibodies (up to six antibodies per candidate). Class I and class II antibodies were then grouped into MFI <5000, MFI 5000–10 000 and MFI >10 000 prior to therapy and analyzed for change over time. The results indicated the greatest decline in MFI was in the 5000–10 000 group with therapy for both class I (p = 0.0075 if incomplete therapy, p = 0.0004 for complete therapy) and class II (p ≤ 0.0001 if complete therapy). However, the estimated change per day of MFI was less than 14. Interestingly, there was no significant decline in the class I or class II MFI <5000, with the exception of those who did not complete therapy where there was a slight decline in class I (p = 0.0007, MFI change 58 per day). In general, the strongest MFI (>10 000 prior to therapy) did not decline with the exception of class I and only those who completed therapy (p ≤ 0.0001).
Tolerability and complications
Eight of 18 who initiated desensitization therapy were able to complete the entire protocol. Reasons for early discontinuation in the other 10 candidates included transplant (n = 3), thrombocytopenia with bleeding concerns (n = 4), supraventricular tachycardia during plasmapheresis (n = 1) and other medical concerns unrelated to the therapy that made the candidate ineligible for transplant (n = 1). An additional candidate was considered to have undergone an incomplete protocol as this candidate was the first to undergo multi-modal desensitization at our center and a shorter, modified protocol was utilized.
Peripheral neuropathy necessitating additional medications was seen in one candidate when bortezomib was given intravenously. Subsequently, the route of administration was changed to subcutaneous and no further peripheral neuropathy was noted.
There were no infections related to therapy.
Subsequent lung transplantation
Of the nine candidates that subsequently received a transplant, two recipients received donors with HLA typing to historical HLA antibodies. Our center uses a virtual cross-match and the most recent sample indicated the virtual cross-match was negative. However, one of these recipients had a positive retrospective cross-match. It is possible that the cross-match was due to therapies given (specifically rituximab) and/or non-HLA antibodies. Four of the nine recipients received donors with HLA typing that did not overlap with any current or historical HLA antibodies. An additional three recipients received donor lungs with HLA typing that matched low MFI level detected antibodies but the prospective cross-match was negative, so the lungs were transplanted.
The survival analysis included all transplant recipients who had pretransplant detectable HLA antibodies transplanted after January 1, 2010. One-year survival was 65% in the HLA positive candidates group undergoing desensitization and 85% in the HLA positive group not desensitized. There was no difference in unadjusted posttransplant 1-year survival between the groups (p = 0.15); however, there were wide confidence intervals in the desensitized group with the small number of recipients (Figure 2). Cause of death within 1 year of transplant in the desensitized group included liver failure immediately postoperative (n = 1), progressive distal airway stenosis (n = 1), anastomotic complication with subsequent multi-organ failure and stroke (n = 1). None of the deaths were thought to be directly related to the desensitization therapy.
An important part of the growth in lung transplant has been through expanded recipient criteria. However, this growth is limited by the available donor pool. As we recognize the fine specificities of HLA antibodies, and in the future have clinical tests available for non-HLA antibody detection, the potential donor pool may further decrease. Desensitization protocols are an important avenue to potentially decrease preformed antibodies prior to transplant. Our multi-modal protocol of plasmapheresis, steroids, rituximab, bortezomib and IVIG did not show a significant decline in class I or II PRA and more importantly, cPRA.
Our findings are in contrast to other reports in the literature demonstrating a benefit with desensitization. There are critical differences in these studies that may partially explain the differing conclusions. First, our cohort included broadly sensitized (combined cPRA ≥ 80%) candidates where other cohorts have been less broadly sensitized. This highly selected group may be quite different in terms of therapy response compared to those with lower percentage PRA. This is supported by at least two other studies in renal transplant with highly sensitized (cPRA levels >90%) candidates not demonstrating an antibody response with rituximab and IVIG [12, 13]. The only other case series in thoracic transplant utilizing multi-modal desensitization therapy (rituximab/IVIG followed by plasmapheresis/bortezomib at a later time) demonstrated a significant decline in a lower starting mean cPRA (62% which decreased to 35%) . Second, it is possible that the current antibody detection methods are confounded by the therapies we utilized. For example, IVIG has been reported to lead to a false positive HLA antibody test as anti-HLA E antibodies in IVIG may cross-react with HLA class I antigens since HLA E antigens share epitopes with HLA class I antigens . Careful consideration of the timing of the test in relation to therapies is critical in assessing response to therapy. Third, advanced organ disease of different organs may alter the antibody production based on the organ type. In an advanced disease state, the lung, with its extensive blood supply and lymphatic tissue, may drive antibody production and protect B cells, while the advanced heart failure or renal failure response to desensitization therapy might be quite different.
A secondary analysis suggested a small, but statistically significant, decrease in MFI for some antibodies. Despite a limited sample analysis and small change in MFI, this is potentially an important finding. Under current clinical practice, only a decline in cPRA would change the donor pool. However, if centers begin to consider lower strength HLA antibodies as acceptable donor-specific antibodies, then desensitization therapies might have a greater benefit and increase the donor pool for a sensitized recipient. A key limitation to our finding is the size of our cohort. While this is largest reported lung transplant candidate cohort undergoing a multi-modal desensitization, it is still a relatively small number of transplant candidates. This protocol is aggressive, expensive and potentially with side effects. We believe an early analysis of the protocol was warranted prior to subjecting more patients to the therapies. A larger cohort may be needed to confirm a significant difference with desensitization.
Given the overall disappointing response of these broadly sensitized candidates to desensitization therapy, other avenues of determining compatible donors should be explored. A critical piece is determining if these antibodies are significant clinically. While we have the ability to detect HLA antibodies, we have limited understanding of their functionality. One avenue under investigation is the ability of the antibody to activate complement. These pathologic antibodies could thus lead to a complement-dependent positive cross-match result and ensuing graft injury. Recently, an assay to distinguish HLA antibodies that bind C1q, the first step in the complement cascade, was developed. The single antigen bead antibody-binding C1q assay correlates with a antibody-mediated rejection . In renal allograft recipients, donor-specific HLA antibodies that bind C1q are associated with graft loss [20, 21]. Similarly in heart transplant recipients, donor-specific HLA antibodies that are C1q+ are associated with CDC positive cross-match and with antibody-mediated rejection [19, 22]. While promising, the C1q assay is in limited clinical use and its broader application for pretransplant lung candidates is uncertain.
Finally, it is important to note that the challenge of HLA antibodies does not end with the transplant. Detailed analysis of pretransplant and posttransplant HLA antibodies demonstrate these antibodies are associated with worse survival in single-center studies while United Network of Organ Sharing registry data analysis suggests no difference in survival for sensitized recipients transplanted after 1998 [23-25]. This underscores the need to better understand the mechanisms of antibodies in transplant candidates and recipients. In summary, an aggressive antibody depletion strategy does not seem to result in lowering of the PRA or cPRA in highly sensitized lung transplant candidates. The lack of a known effective strategy for optimizing these patients both pretransplant and posttransplant poses a major hurdle to extending the life-saving benefits of transplant to more people in need.
The authors would like to acknowledge the statistical support and guidance of Dr. Barry Moser, Department of Biostatistics and Bioinformatics, Duke University School of Medicine. Funding source was Duke University School of Medicine.
The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.