This was a single-center prospective observational study performed under the approval of the Institutional Review Board of the University of Pittsburgh (STUDY 19090098 approval: 12/2019; ClinicalTrials.gov Identifier: NCT04318587). Adults (>18 years-old, both male and female) undergoing de-novo lung transplantation at our institute were recruited in this study. The patients who had multi-organ transplants, bone-marrow transplants, and re-transplantation were excluded due to the detection mechanism of dd-cfDNA. All subjects were consented to the study before the transplantation. The blood samples from the consented subjects were obtained using a blood tube preserving the cfDNA (Cell-Free DNA BCT®; Streck Corporate, La Vista, NE) and scheduled to draw at pre-operation and postoperative day 1, 2, 3, once a week for 1–12 weeks, and once a month from 4 months to 12 months posttransplant (Figure 1A). After the subjects were discharged, the blood was drawn at the time of their follow-up visit.

The dd-cfDNA were measured using the AlloSure assay (CareDx, Inc., Brisbane, CA). Allosure assay is a clinical-grade, next generation of targeted gene sequencing assay to measure SNPs to accurately quantify dd-cfDNA. All blood samples were shipped to the CareDx facility for the AlloSure assay. The dd-cfDNA data were provided as a percentage (%dd-cfDNA) of the total (donor + recipient) cfDNA.

The primary outcomes include the PGD grade, acute cellular rejection (ACR), and development of de-novo donor-specific antibody (DSA). The PGD grade for the first posttransplant 3 days was determined daily by the treating team, using the ISHLT criteria (15). Our protocol includes a surveillance bronchoscopy with TBBx at 2–3 weeks post-transplant. ACR was diagnosed and graded by a pulmonary pathologist after evaluation of histopathology slides (16).

After the subject discharging, allograft ACR was diagnosed by TBBx at the time of the subject's follow-up visit and was graded. At the same time as TBBx, blood was sampled for ddcfDNA for AlloSure assay and detecting DSA by the UPMC Histocompatibility Laboratory (17). Positive DSA was determined with mean fluorescence intensity >1,000 in a single-antigen beads assay (LABScreen class I and II, One Lambda, West Hills, CA). We analyzed the %dd-cfDNA data to see the correlation with the ACR grade and the existence of de-novo DSA at the same time point.

Univariable analyses were performed using SPSS (v. 28.0, IBM, Armonk, NY). Results are shown as mean ± standard deviation (SD) with individual values or median and interquartile range (IQR). Univariable comparisons were performed using chi-square tests for categorical variables and Mann-Whitney U tests for continuous variables. Data with multiple groups were analyzed using one-way analysis of variance followed by post hoc analysis with the Tukey test for multiple comparisons. Multiple observations over time were analyzed using a two-way ANOVA mixed model. A probability level of p < 0.05 was considered statistically significant.

The patient characteristics and posttransplant data are summarized in Table 1. Forty patients (39 double lung transplants and 1 single lung transplant) were enrolled. All recipients were transplanted with grafts from brain death donors. 39 sequential double lung transplants and 1 single lung transplant were performed. One case used ex-vivo lung perfusion to evaluate a graft before transplant. The severe PGD incidence at 72 h posttransplant was 27.5% (n = 11). The mean duration of intensive care unit and hospital stay were 10.7 ± 12.9 and 27.3 ± 16.2 days, respectively (Table 2).

A total of 515 samples were collected to determine %dd-cfDNA from the subjects. Twenty-four samples were rejected for the assay due to poor sample quality or the assay metrics issue. The time-dependent change of %dd-cfDNA in the first 3 months is shown in Figure 1B. The %dd-cfDNA in the circulation was elevated from baseline immediately after transplantation (pretransplant: mean ± SD 0.013 ± 0.05%) and showed a peak value (mean ± SD 15.5 ± 4.56%, range: 5.94–26.99%) within 48 h posttransplant. The %dd-cfDNA was decreased over 2 weeks and reached a plateau (stable) level (Figure 1B). There was no significant association between the lung allograft preservation time and the peak %dd-cfDNA (Figure 1C).

We compared the circulating %dd-cfDNA between patients who developed PGD grade 3 and those who with grade 0–2, with the hypothesis that lungs with severe PGD should exhibit a greater degree of allograft cellular death, releasing a higher plasma level of dd-cfDNA. Instead, we found a lower %dd-cfDNA at T48 h in the severe PGD cohort, and no difference in %dd-cfDNA at any other timepoints up to 4 weeks (Figure 2A). We next found no difference in the %dd-cfDNA based on grade of PGD (Figure 2B). Finally, we found no statistically significant association between the peak %dd-cfDNA with the PGD3 within 24–48 h posttransplant (Figure 2C).

At our institution, we perform the first TBBx within 2–3 weeks of transplantation to detect early ACR. We compared the %dd-cfDNA in the days to weeks after transplantation with the histopathological diagnosis of early ACR. When looking at longitudinal, timepoint matched values of circulating %dd-cfDNA, we found no statistically significant difference between those study participants who developed early ACR and those that did not (Figure 3A). However, we found a statistically significant increase in %dd-cfDNA in the plasma drawn from recipients posttransplant 2 weeks and the first TBBx showing early ACR (Figure 3B). Although statistically significant, there was substantial overlap between groups, and the upper limit of %dd-cfDNA in the early ACR group was well below levels found at 1 week after transplantation.

We next looked at the %dd-cfDNA compared to the presence of ACR and de-novo DSA diagnosed after discharging to 1-year follow-up period. We analyzed data to see the correlation of %dd-cfDNA with the ACR grade based on TBBx and the existence of de-novo DSA at the same time point. Contrary to recent reports about the association of antibody mediated rejection and %dd-cfDNA (12), we found that %dd-cfDNA were not significantly higher in patients with de-novo DSA (Figure 3C). However, patients with both ACR and de-novo DSA had a significantly higher %dd-cfDNA than those without either (Figure 3C). Interestingly, other sources of allograft injury (bronchiolitis, sepsis, anastomosis stenosis, and pseudomonas pneumonia) had comparable elevation in %dd-cfDNA as those with both ACR and de-novo DSA. These findings suggest that %dd-cfDNA could not differentiate sources of perioperative lung injury.

We provisionally investigated the association of %dd-cfDNA determined by the next generation of gene sequence technology with early perioperative outcomes including the PGD, ACR, and de-novo DSA in recipients of lung transplantation. Regardless of PGD grade, most study participants had a substantial increase in the proportion of dd-cfDNA immediately after lung transplantation. Similar results have been observed in kidney transplantation and this is considered a result of ischemia-reperfusion injury (18). While we observed the %dd-cfDNA reached the baseline level within 2 weeks and decreased at 86.5% from the peak value in this study, the trend of %dd-cfDNA could differ among the organ types. Studies in kidney transplant demonstrated that %dd-cfDNA converged to the plateau within a week (19–21), while the multicenter study “GRAfT” for heart transplants showed it reached the baseline by ∼60 days (22). Allograft cell death should correlate with grade of ischemia-reperfusion injury and PGD after solid organ transplantation (23). Our findings in lung transplant setting are not consistent with this hypothesis owing to 3 potential explanations. First is the nature of the test itself, measuring the proportion and not absolute value of the dd-cfDNA; second, is the systemic symptom of PGD3; and the third, is the unknown source of the cfDNA. As previously mentioned, the %dd-cfDNA is the ratio of dd-cfDNA among the total cfDNA circulating in the blood of recipients which is likely impacted by recipient-derived cfDNA, especially with inflammatory cell turnover. Thus, increased occupation of plasma recipient-derived cfDNA in response to systemic conditions can reduce the %dd-cfDNA proportionally (24). While %dd-cfDNA is often utilized in clinical currently rather than the absolute copy or concentration of cfDNA, the quantification of dd-cfDNA copy may represent the PGD phenotype or rejection without recipients' noise (25, 26). Since this method measures the ratio, there is no method to normalize across patient characteristics or procedure-related biases (e.g., body size, patient condition, single/double lung transplant, blood transfusion).

Severe PGD is now appreciated to be a systemic condition which may impact the proportion of donor and recipient derived cfDNA (27). Severest PGD patients are often comorbid with acute kidney failure or liver dysfunction and these lead patients to systemic illness and require additional life support therapy (28, 29). Some of our subjects with severe PGD showed critical illness in this study and required hemodialysis in addition to extracorporeal membrane oxygenation (ECMO). These factors could affect both recipient and total DNA levels and create potential bias on %dd-cfDNA through material-biological interaction, increased recipient-derived cfDNA, and elimination of DNA by hemodialysis (30–32). Further basic investigations are required to clarify the effects of these disturbances on the net cfDNA and the diagnosis power of dd-cfDNA for posttransplant graft outcomes in lung transplantation.

Another important aspect of our finding is that the spike of the %dd-cfDNA was observed in the non-PGD group as well, suggesting this cellular death is not associated with graft injury and dysfunction. undetectable cellular source of cfDNA is one of the limitations of the bulk dd-cfDNA assay, thus we could not know whether the dd-cfDNA was resulting from the tissue injury after ischemia-reperfusion, or just washing out effects of the remnant cfDNA carried over from the donors in grafts. While the vascular bed of grafts is washed by antegrade and retrograde flushing at procurement, some donor cells [aka. microchimerism (33)] and other blood components could remain in the allografts (23, 34). Therefore, the dd-cfDNA released from the allografts immediately after transplantation could contain multiple factors, and determining the source of the dd-cfDNA within the first 3 days is of interest to define its association with the allograft damage and increase the graft specificity (35).

We observed significantly increased %dd-cfDNA was associated with ACR alone and combined allograft rejection (ACR + de-novo DSA), but not de-novo DSA alone. Despite these statistically significant increases in %dd-cfDNA with both ACR and combined rejection, the percentage point increases were small; the majority had less than 2% in most cases. This modest increase in %dd-cfDNA does not provide us much insight into the relationship between graft cell death and the alloreactive response. It is entirely possible that rate of graft cell death occurs much more slowly in the setting of allorecognition (36). Alternatively, the low proportion of %dd-cfDNA in the setting of rejection represents a concomitant increase in circulating cfDNA of both donor graft cell death and recipient-derived immune cell turnover, leading to variable and modest changes in the circulating %dd-cfDNA. As mentioned, the lifetime of cfDNA in blood circulation is usually 1–2 h, thus it is hard to capture a slow-progressing cellular death in grafts by detecting dd-cfDNA. Multi-factorial rejection may enhance cell death in grafts thus the %dd-cfDNA has responded higher in the mixed allograft damage than the single factor rejection.

Ultimately, the limited range of circulating %dd-cfDNA in the setting of early allograft rejection raises questions about the ability of circulating %dd-cfDNA to identify specific patterns of allograft injury with any degree of resolution and sensitivity. This poor resolution as a diagnostic tool has been reported in other solid organ transplants, with levels range above 1.0% of dd-cfDNA for a risk of active rejection, especially above 0.5% for lung transplantation (37). Not only the graft rejection, the %dd-cfDNA in lung transplantation could correlate with exogenous and endogenous effectors, such as aspiration, bronchiolitis, infection, or malignancy, which augment mass cell death. Thus, the %dd-cfDNA can supplementally be important to support diagnosing graft rejection with TBBx, DSA detection, or other conventional tools. More sophisticated means of determining the cellular origin of circulating cfDNA used in concert with other metrices may provide the detection of dd-cfDNA the granularity needed to be a clinical useful biomarker. However, as it remains now, the %dd-cfDNA does not provide the precision necessary to justify broad clinical utilization.

There are several limitations of dd-cfDNA study in general. Without performing the donor genotyping, the next generation of gene sequencing technology cannot identify the dd-cfDNA in the patient plasma after re-transplantation, multi-organ transplantation, and bone marrow Tx case. In addition, we were not able to get the %dd-cfDNA data for 24 samples out of 515 samples, because of the sample quality and the assay quality metrics issue. One of the biggest issues in the sample quality is hemolysis and hemolysis can happen with improper sampling in general or a longer sample storage in the tube [>5 days (38)], or it may be originated in the patients due to ECMO, cardiopulmonary bypass, blood transfusion or hemodialysis. This may result in an increase in the background recipient-derived cfDNA, thus those samples were rejected from the assay (39).