Construct validity of the definition of primary graft dysfunction after lung transplantation

Article Outline

• Abstract
• Methods
  • Study design and populations
    • Biomarkers
    • Survival
  • Definition of PGD
  • Analysis of varying definitions of PGD on association of biomarkers of acute lung injury
  • Impact of varying PGD definitions on mortality
• Results
• Discussion
• Disclosure statement
• Appendix
  • Participants in the Lung Transplant Outcomes Group by Site
• References
• Copyright

Background

This study tested the discriminant validity of International Society for Heart and Lung Transplantation (ISHLT) primary graft dysfunction (PGD) grades with lung injury biomarker profiles and survival.

Methods

The study samples consisted of a multicenter prospective cohort study for the biomarker analysis and a cohort study of 450 patients for the mortality analyses. PGD was defined according to ISHLT consensus at 24, 48, and 72 hours after transplantation. We compared the changes in plasma markers of acute lung injury between PGD grades using longitudinal data models. To test predictive validity, we compared differences in the 30-day mortality and long-term survival according to PGD grade.

Results

PGD Grade 3 demonstrated greater differences between plasma intercellular adhesion molecule 1 (ICAM-1), protein C, and plasminogen activator inhibitor type 1 (PAI-1) levels than did PGD Grades 0 to 2 at 24, 48, and 72 hours after lung transplantation (p < 0.05 for each). Grade 3 had the highest 30-day (test for trend p < 0.001) and overall mortality (log rank p < 0.001), with PGD Grades 1 and 2 demonstrating intermediate risks of mortality. The ability to discriminate both 30-day and overall mortality improved as the time of grading moved away from the time of transplantation (test for trend p < 0.001).

Conclusions

The ISHLT grading system has good discriminant validity, based on plasma markers of lung injury and mortality. Grade 3 PGD was associated with the most severely altered plasma biomarker profile and the worst outcomes, regardless of the time point of grading. PGD grade at 48 and 72 hours discriminated mortality better than PGD grade at 24 hours.

Keywords: lung transplantation, complications, acute lung injury, primary graft dysfunction, reperfusion injury

Primary graft dysfunction (PGD) is an acute lung injury syndrome that occurs in the post-transplant period and is characterized by radiographic pulmonary infiltrates and hypoxemia. Clinically and pathologically, the syndrome is similar to acute respiratory distress syndrome (ARDS). PGD represents a spectrum of injury from mild to more severe hypoxemia and lung injury. The more severe forms of PGD have been associated with worse morbidity, and PGD is the leading cause of death in the early post-transplant period.¹, ², ³, ⁴, ⁵

PGD research has been hampered by the lack of a consistent definition or gold standard.⁶, ⁷, ⁸, ⁹ For example, different criteria have led to mortality estimates for PGD of 21% to 63%, as well as inconsistent reporting of clinical risk factors.¹, ¹⁰, ¹¹, ¹² In 2005, the International Society for Heart and Lung Transplantation (ISHLT) published a consensus statement with the aim of standardizing the definition and grading of PGD.¹, ⁶, ⁷, ¹³, ¹⁴ Similar to that of ARDS, the grading scheme for PGD considers 2 factors: the appearance of post-transplant chest X-ray images and the ratio of the partial pressure of arterial oxygen to the fraction of inspired oxygen (Pao₂/FiO₂) at multiple time points within the first 72 hours after transplantation.⁶, ¹⁵

This study proposed to assess the optimal grading and timing of the definition of PGD for use in clinical and translational studies aimed at elucidating mechanisms of and risks for PGD.⁶, ⁹ We compared the association between established plasma biomarkers of acute lung injury with PGD variously defined, and we tested the association of PGD definitions with mortality risk.

Back to Article Outline

Methods 

The research protocol of this study was approved by the Institutional Review Boards at each of the participating centers.

Study design and populations 

Biomarkers 

We performed a prospective cohort study of patients undergoing first lung transplantation at 6 centers in the United States participating in the Lung Transplant Outcomes Group (LTOG; see Appendix for institutions and investigators.) The study sample consisted of 128 adult lung transplant recipients enrolled between June 2003 and November 2004 at these centers, with biomarker results published previously.¹⁶, ¹⁷

Survival 

We performed a retrospective cohort study of 450 consecutive adult lung transplant procedures at the University of Pennsylvania between October 1991 and October 2005. The follow-up period for survival analysis extended to October 2009. We excluded 2 heart-lung recipients and 2 lung-liver recipients. Thus, the study sample comprised 446 patients.

Definition of PGD 

We used the ISHLT consensus definition of PGD.¹⁸ Specifically, PGD grade was based on the following criteria:

All patients receiving extracorporeal membrane oxygenation were classified as Grade 3 PGD. All patients receiving oxygen by nasal cannula with FiO₂ estimated as less than 0.3 were Grade 0 or 1, based on chest X-ray findings. These criteria were applied at the 24, 48, and 72-hour time points (T24, T48, and T72) after transplantation according to ISHLT guidelines.¹⁸

Analysis of varying definitions of PGD on association of biomarkers of acute lung injury 

Our hypothesis was that higher grades of PGD would have different levels of biomarkers of acute lung injury than lower grades. To address this hypothesis, we chose established biomarkers of lung injury from prior studies: protein C, type 1 plasminogen activator inhibitor (PAI-1), and type 1 Intracellular adhesion molecule (ICAM-1).¹⁶, ¹⁷, ¹⁹, ²⁰ We measured plasma levels of these biomarkers at 6, 24, 48, and 72 hours after reperfusion in 128 lung transplant patients enrolled in the LTOG; the main results have been presented in prior publications.¹⁶, ²¹ Standard enzyme-linked immunosorbent assay methods were used and are described elsewhere.¹⁶, ²¹

Logistic regression models were fit and stratified by biomarker measurement time point and PGD grading day for each biomarker separately. Because there were 2 time dimensions to consider in these analyses (e.g., biomarker assessment time and PGD grading time), the stratified analyses fixed both time dimensions, and summaries were presented for each biomarker measurement time (baseline, Time 1, Time 2, and Time 3) and PGD grading day combination (grading on Day 1, Day 2, and Day 3). These models were fit to estimate the association of each biomarker on PGD and non-PGD patients using a range of grading criteria to define PGD status (e.g., grade = 3 vs grade = 2 or 1 or 0; grade = 2 or 3 vs grade = 1 or 0; grade = 3 or 2 or 1 vs grade = 0). Adjusted for biomarker, the predicted probability of PGD status and estimated risk differences (estimated probability of PGD case minus estimated probability of PGD control) with corresponding 95% confidence intervals are presented to estimate the influence of each biomarker in predicting PGD status. When adjusting for biomarker in the fitted models, we standardized each by their respective estimated standard deviation.

Impact of varying PGD definitions on mortality 

We compared both 30-day mortality and overall survival between the different grades of PGD measured at different time points. Overall survival was assessed using Kaplan-Meier methods with log-rank tests. Cox proportional hazards models were constructed to compare hazard ratios between PGD grades at different time points. To estimate differences in the effect of PGD on mortality according to transplant procedure type, multiplicative interaction terms were used in logistic regression models (for 30-day mortality) and Cox models (for overall survival).

All statistical comparisons were performed using STATA 10.1 software (STATA Corp, College Station, TX), and SAS 9.1 software (SAS Institute, Cary, NC).

Back to Article Outline

Results 

Table 1 summarizes the characteristics of the study sample for the biomarkers analysis.¹⁹, ²² The mean age of recipients was 50 years (95% CI, 47–53 years), 45% were women, 88% were Caucasian, and the most common predisposing diagnosis was chronic obstructive pulmonary disease. The clinical characteristics of the LTOG biomarker population according to PGD Grade 3 status have been published previously.¹⁶, ²¹

Table 1. Clinical Characteristics of the Study Populations

CI, confidence interval; COPD, chronic obstructive pulmonary disease.

Differences in levels of protein C, ICAM-1, and PAI-1 were greatest for PGD Grade 3 (at each time point) vs PGD Grades 0–2, supporting the higher severity of the PGD Grade 3 (Figure 1). Even at the earliest time points, the levels of each biomarker were significantly different for PGD Grade 3 compared with the other grades and remained so throughout the assessed time points. Table 2 reports the corresponding risk differences of PGD (with varying definitions) conferred by plasma biomarkers at the 4 measurement times. Risk differences represent the difference in the risk of PGD (however defined) between patients with any value of biomarker compared with patients with a biomarker value that is 1 standardized unit higher. Across most of the post-operative assessments, analyses of Grade 3 vs Grades 0, 1 and 2 produced the greatest biomarker discrimination, as evidenced by the greater risk differences. In certain instances, defining PGD as Grades 2 or 3 vs Grades 0 or 1 produced similar risk difference estimates. Although ICAM-1 and protein C produced the best discrimination when PGD was graded at 48 or 72 hours, PAI-1 showed greater differences at 24 hours.

• 
  • View Large Image
  • Download to PowerPoint

• Figure 1. 

  Biomarkers of acute lung injury by different primary graft dysfunction (PGD) grades at 72 hours (T72): (A) protein C, (B) plasminogen activator inhibitor type 1 (PAI-1), (C) intercellular adhesion molecule 1 (ICAM-1). The different grades appear as numbers in the lines at the different time points. Solid lines indicate Grade 0; dotted lines indicate Grade 1; dashed lines indicate Grade 2; and dash-dot lines indicate Grade 3 PGD. *p < 0.05; **p < 0.01 for Grade 3 vs the rest.

Table 2. Biomarker Discrimination Summaries of Patients With and Without Primary Graft Dysfunction, as Defined by Various Grade Thresholds, by Biomarker Measurement Time^a

ICAM-1, intercellular adhesion molecule 1; PAI-1, plasminogen activator inhibitor type 1.

aData are expressed as risk differences and 95% confidence intervals.

Table 1 summarizes the characteristics of the study sample in the survival analysis. The mean age of recipients was 53 years (95% CI, 52–54 years), 43% were women, and 89% were Caucasian. The overall prevalence of each PGD grade according to different time points is presented in Table 3. Overall, the prevalence of higher grades of PGD significantly declined farther out from the time of transplant (test for trend p < 0.001). For example, the prevalence of PGD Grade 3 was 28% at T24 and 18% at T72. On the other hand, 33% had PGD Grade 0 (no PGD) at T0, whereas 47% had no PGD by T72.

Table 3. Prevalence (with 95% confidence intervals) of Different Grades of Primary Graft Dysfunction at Different Time Points in the Outcomes Study

PGD, primary graft dysfunction.

PGD grades at different times were associated with 30-day mortality (Figure 2). Using PGD grade at T24, 30-day all-cause mortality was 24.5% (95% CI 17.2–33.2) for Grade 3, 6.2% (95% CI 2.1–13.9) for Grade 2, 3.5% (95% CI 0.7–9.8) for Grade 1, and 4.2% (95% CI 1.5–9.0) for Grade 0. According to PGD grade at T72, 30-day all-cause mortality was 36.4% (95% CI 25.7–48.1) for Grade 3, 6.1% (95% CI 1.2, 16.9) for Grade 2, 5.0% (95% CI 1.6–11.2) for Grade 1, and 3.5% (95% CI 1.4–7.0) for Grade 0 (test for trend, p < 0.001). There was no interaction between PGD grade and the type of transplant (single vs bilateral) at any time point (all p > 0.50), indicating that the impact of PGD on 30-day mortality did not vary by the type of transplant performed.

• 
  • View Large Image
  • Download to PowerPoint

• Figure 2. 

  Impact of primary graft dysfunction grades at different time points on 30-day mortality rates.

Figure 3 presents differences in overall survival by PGD grade. The overall all-cause mortality was significantly different between the groups (p < 0.001 by log-rank test for each time point). Although Grade 3 PGD at T24 had a significant worse outcome compared with other PGD grades, Grade 3 compared with other PGD grades at T72 demonstrated more pronounced differences in survival. Likewise, at times further from transplant, intermediate grades of lung injury (Grades 1 and 2) discriminated mortality differently from Grade 0 and Grade 3. There was no interaction between PGD grade and type of transplant (single vs bilateral) at any time point (p > 0.50 for each time point), indicating that the impact of PGD on overall survival did not vary by the type of transplant performed.

• 
  • View Large Image
  • Download to PowerPoint

• Figure 3. 

  Kaplan-Meier survival curves of PGD grades at different time points: (A) 24 hours (T24), (B) 48 hours (T48), and (C) 72 hours (T72).

Back to Article Outline

Discussion 

In this study, we demonstrated the discriminant validity of the ISHLT grading system for PGD for biomarker profiles and survival. PGD Grade 3 showed the most significant alterations in plasma biomarker profiles of lung injury. Grade 3 was also associated with the highest risk of death at 30 days and worse long-term survival. Grading PGD at times farther out from transplant demonstrated sharper discrimination of both early mortality and long-term survival. These results may inform scientific definitions of PGD to be used in future studies aimed at investigating PGD risk and outcomes.

PGD Grade 3 appeared distinct from Grades 0 to 2 in biomarker profile and survival. Therefore, studies of mechanisms, risk factors, or prevention of ischemia-reperfusion injury after lung transplantation may focus on PGD Grade 3, given the plasma biomarker differences as well as the significantly increased risk of short-term and long-term mortality. These results support the use of either “any PGD Grade 3” or “PGD Grade 3 at T72” as the primary outcome for mechanistic studies seeking a dichotomous PGD definition¹², ²³, ²⁴ and in future clinical trials of treatment or prevention. Nonetheless, given that PDG Grade 2 produced similar biomarker discrimination at some measurement time points, investigators may choose to account for this apparent intermediate injury by treating PGD as an ordinal variable or by excluding Grade 2 individuals.²², ²⁵

We found that as PGD grading was applied at later times, discrimination for mortality improved. The better discrimination of PGD grade farther out from transplant for mortality may have been due to misclassification of recipients with early, mild, reversible pulmonary edema. Alternately, the better discrimination of PGD grade at later times may represent persistent or irreversible lung injury that may have greater impact on outcome. Grading earlier after transplantation demonstrated a significantly higher prevalence of more severe PGD, reflecting either a higher risk of early acute lung injury or a distinct lung injury phenotype.

PGD Grades 1 and 2 were associated with a higher risk of death compared with PGD Grade 0 that appeared to manifest after the first 30 days. These findings are consistent with those of Daud et al,²³ who demonstrated a stepwise association of PGD grade with the development of bronchiolitis obliterans syndrome. That mild and moderate lung injury is associated with later chronic rejection is consistent with the injury response hypothesis²⁶ and warrants further investigation. Likewise, the absence of clinically apparent lung injury (PGD Grade 0) at any time was associated with the best long-term survival.

Our study had several limitations. First, we did not perform grading at T0.¹⁵, ²⁷, ²⁸ The lack of grading for earlier time points may have hampered our ability to demonstrate differences in PGD grade with mortality by different transplant procedure types.⁸ However, we did not detect a difference when testing for interaction by transplant type in our analyses of biomarkers or outcomes. These findings can be interpreted as consistent with Oto et al,⁸ who found the greatest differences between transplant type to be at earlier grading time points, whereas later grading seemed to give similar results between transplant types. In addition, the potential utility of earlier grading time points is supported by our observation that the measured plasma biomarkers exhibited spikes at the earliest measured time point (6 hours). Finally, we only assessed short-term and long-term mortality. Other important end points include acute rejection and bronchiolitis obliterans syndrome, which previously have been linked with early allograft injury in a step-wise fashion according to PGD grade.¹², ²³ Future studies will focus on these different outcomes in explaining mortality differences.

We conclude that the ISHLT consensus PGD definition demonstrates evidence of convergent and divergent validity using mortality and concurrent biomarkers as constructs. PGD Grade 3 after lung transplantation appears different from other PGD grades, regardless of time point. Our results suggest PGD Grade 3 is a useful dichotomous definition for studies aimed at PGD mechanism or PGD prevention. In addition, our results support the specification of intermediate PGD grades for long-term outcome studies. Future studies using these definition of PGD will elucidate the mechanism and possible interventions to reduce the risk of this complication and potentially improve long-term outcomes of lung transplantation.

Back to Article Outline

Disclosure statement 

This study was funded by National Institutes of Health (NIH) HL04243, HL081619, HL087115, HL67771, HL081332, HL088263, and the Craig and Elaine Dobbin Pulmonary Research Fund.

The authors have no relationships with commercial entities that have an interest in the subject matter of this manuscript.

Back to Article Outline

Appendix 

Participants in the Lung Transplant Outcomes Group by Site 

Back to Article Outline

References 

1. Arcasoy SM, Fisher A, Hachem RR, et al. Report of the ISHLT Working Group on Primary Lung Graft Dysfunction part V: predictors and outcomes. J Heart Lung Transplant. 2005;24:1483–1488
   • View In Article
   • Full Text
   • Full-Text PDF (117 KB)
   • CrossRef
2. Christie JD, Bavaria JE, Palevsky HI, et al. Primary Graft Failure following lung transplantation. Chest. 1998;114:51–60
   • View In Article
   • MEDLINE
   • CrossRef
3. Christie JD, Kotloff RM, Ahya VN, et al. The effect of primary graft dysfunction on survival after lung transplantation. Am J Respir Criti Care Med. 2005;171:1312–1316
   • View In Article
4. Christie JD, Kotloff RM, Pochettino A, et al. Clinical risk factors for primary graft failure following lung transplantation. Chest. 2003;124:1232–1241
   • View In Article
   • MEDLINE
   • CrossRef
5. King RC, Binns OA, Rodriguez F, et al. Reperfusion injury significantly impacts clinical outcome after pulmonary transplantation. Ann Thorac Surgery. 2000;69:1681–1685
   • View In Article
6. Christie JD, Carby M, Bag R, et al. Report of the ISHLT Working Group on Primary Lung Graft Dysfunction part II: definition (A consensus statement of the International Society for Heart and Lung Transplantation). J Heart Lung Transplant. 2005;24:1454–1459
   • View In Article
   • Full Text
   • Full-Text PDF (93 KB)
   • CrossRef
7. Christie JD, Van Raemdonck D, de Perrot M, et al. Report of the ISHLT Working Group on Primary Lung Graft Dysfunction part I: introduction and methods. J Heart Lung Transplant. 2005;24:1451–1453
   • View In Article
   • Full Text
   • Full-Text PDF (51 KB)
   • CrossRef
8. Oto T, Levvey BJ, Snell GI. Potential refinements of the International Society for Heart and Lung Transplantation primary graft dysfunction grading system. J Heart Lung Transplant. 2007;26:431–436
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (316 KB)
   • CrossRef
9. Zaas D, Palmer SM. Respiratory failure after lung transplantation: now that we know the extent of the problem, what are the solutions?. Chest. 2003;123:14–16
   • View In Article
   • MEDLINE
   • CrossRef
10. Chatilla WM, Furukawa S, Gaughan JP, et al. Respiratory failure after lung transplantation. Chest. 2003;123:165–173
    • View In Article
    • MEDLINE
    • CrossRef
11. Thabut G, Vinatier I, Stern JB, et al. Primary graft failure following lung transplantation: predicitive factors of mortality. Chest. 2002;121:1876–1882
    • View In Article
    • MEDLINE
    • CrossRef
12. Whitson BA, Prekker ME, Herrington CS, et al. Primary graft dysfunction and long-term pulmonary function after lung transplantation. J Heart Lung Transplant. 2007;26:1004–1011
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (500 KB)
    • CrossRef
13. Barr ML, Kawut SM, Whelan TP, et al. Report of the ISHLT Working Group on Primary Lung Graft Dysfunction part IV: recipient-related risk factors and markers. J Heart Lung Transplant. 2005;24:1468–1482
    • View In Article
    • Full Text
    • Full-Text PDF (189 KB)
    • CrossRef
14. de Perrot M, Bonser RS, Dark J, et al. Report of the ISHLT Working Group on Primary Lung Graft Dysfunction part III: donor-related risk factors and markers. J Heart Lung Transplant. 2005;24:1460–1467
    • View In Article
    • Full Text
    • Full-Text PDF (105 KB)
    • CrossRef
15. Oto T, Levvey BJ, Snell GI. Potential refinements of the International Society for Heart and Lung Transplantation primary graft dysfunction grading system. J Heart Lung Transplant. 2007;26:431–436
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (316 KB)
    • CrossRef
16. Christie JD, Robinson N, Ware LB, et al. Association of protein C and type 1 plasminogen activator inhibitor with primary graft dysfunction. Am J Respir Crit Care Med. 2007;175:69–74
    • View In Article
    • CrossRef
17. Covarrubias M, Ware LB, Kawut SM, et al. Plasma intercellular adhesion molecule-1 and von Willebrand factor in primary graft dysfunction after lung transplantation. Am J Transplant. 2007;7:2573–2578
    • View In Article
    • CrossRef
18. Christie JD, Carby M, Bag R, et al. Report of the ISHLT Working Group on Primary Lung Graft Dysfunction part II: definition (A consensus statement of the International Society for Heart and Lung Transplantation). J Heart Lung Transplant. 2005;24:1454–1459
    • View In Article
    • Full Text
    • Full-Text PDF (93 KB)
    • CrossRef
19. Ware LB, Fang X, Matthay MA. Protein C and thrombomodulin in human acute lung injury. Am J Physiol Lung Cell Mol Physiol. 2003;285:L514–L521
    • View In Article
    • MEDLINE
20. Flori HR, Ware LB, Glidden D, et al. Early elevation of plasma soluble intercellular adhesion molecule-1 in pediatric acute lung injury identifies patients at increased risk of death and prolonged mechanical ventilation. Pediatr Crit Care Med. 2003;4:315–321
    • View In Article
    • MEDLINE
    • CrossRef
21. Covarrubias M, Ware LB, Kawut SM, et al. Plasma intercellular adhesion molecule-1 and von Willebrand factor in primary graft dysfunction after lung transplantation. Am J Transplant. 2007;7:2573–2578
    • View In Article
    • CrossRef
22. Plomin R, Haworth CMA, Davis OSP. Common disorders are quantitative traits. Nat Rev Genet. 2009;10:872–878
    • View In Article
    • CrossRef
23. Daud SA, Yusen RD, Meyers BF, et al. Impact of immediate primary lung allograft dysfunction on bronchiolitis obliterans syndrome. Am J Respir Crit Care Med. 2007;175:507–513
    • View In Article
    • CrossRef
24. Whitson BA, Nath DS, Johnson AC, et al. Risk factors for primary graft dysfunction after lung transplantation. J Thorac Cardiovasc Surg. 2006;131:73–80
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (185 KB)
    • CrossRef
25. Ransohoff DF, Feinstein AR. Problems of spectrum and bias in evaluating the efficacy of diagnostic tests. N Engl J Med. 1978;299:926–930
    • View In Article
    • MEDLINE
    • CrossRef
26. Halloran PF, Homik J, Goes N, et al. The “injury response”: a concept linking nonspecific injury, acute rejection, and long-term transplant outcomes. Transplant Proc. 1997;29:79–81
    • View In Article
    • Full-Text PDF (337 KB)
    • CrossRef
27. Prekker ME, Herrington CS, Hertz MI, et al. Early Trends in PaO(2)/fraction of inspired oxygen ratio predict outcome in lung transplant recipients with severe primary graft dysfunction. Chest. 2007;132:991–997
    • View In Article
    • CrossRef
28. Sekine Y, Waddell TK, Matte-Martyn A, et al. Risk quantification of early outcome after lung transplantation: donor, recipient, operative, and post-transplant parameters. J Heart Lung Transplant. 2004;23:96–104
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (133 KB)
    • CrossRef