- Email: [email protected]
Validation of the Proposed International Society for Heart and Lung Transplantation Grading System for Primary Graft Dysfunction After Lung Transplantation
CLINICAL LUNG AND HEART/LUNG TRANSPLANTATION
Validation of the Proposed International Society for Heart and Lung Transplantation Grading System for Primary Graft Dysfunction After Lung Transplantation Matthew E. Prekker, MD, D. S. Nath, MD, A. R. Walker, MD, A. C. Johnson, MD, M. I. Hertz, MD, C. S. Herrington, MD, D. M. Radosevich, PhD, and Peter S. Dahlberg, MD, PhD Background: A scoring system was recently proposed to grade the severity of primary graft dysfunction (PGD), a frequent early complication of lung transplantation. The purposes of this study are to: (1) validate the PGD grading system with respect to patient outcomes; and (2) compare the performance of criteria employing the arterial oxygenation to fraction of inspired oxygen (P/F) ratio to an alternative grading system employing the oxygenation index (OI). Methods: We retrospectively reviewed the medical records of 402 patients having undergone lung transplantation at our institution from 1992 through 2004. The ISHLT PGD grading system was modified and grades were assigned up to 48 hours post-transplantation as follows: Grade 1 PGD, P/F ⬎300; Grade 2, P/F 200 to 300; and Grade 3, P/F ⬍200. A worst score T(0 – 48) was also assigned, which reflects the highest grade recorded between T0 and T48. Results: The prevalence of severe PGD (P/F Grade 3) declined after transplant, from 25% at T0 to 15% at T48. Grouping patients by P/F grade at T48 demonstrated the clearest differentiation of 90-day death rates (Grade 1, 7%; Grade 2, 12%; Grade 3, 33%) (p ⫽ 0.0001). T48 OI grade also differentiates 90-day death rates. There was no difference in longer-term survival between patients with PGD Grades 1 and 2. OI grade at T0 qualitatively improved differential mortality between Grades 1 and 2; however, the differences did not reach statistical significance. Patients with a worst score T(0 – 48) of Grade 3 PGD did have significantly decreased long-term survival, as well as longer ICU and hospital stay, when compared with Grades 1 and 2 PGD. Significant risk factors for short- and long-term mortality in our multivariate model were P/F Grade 3 [worst score T(0 – 48) as well as T0 grade], single-lung transplant, use of cardiopulmonary bypass and high pre-operative mean pulmonary artery pressure. Conclusions: There is an increased risk of short- and long-term mortality and length of hospital stay associated with severe (Grade 3) PGD. The proposed ISHLT grading system can rapidly identify patients with poor outcomes who may benefit from early, aggressive treatment. Refinement of the scoring system may further improve patient risk stratification. J Heart Lung Transplant 2006;25:371– 8. Copyright © 2006 by the International Society for Heart and Lung Transplantation.
Transplantation is the only effective treatment for several end-stage lung diseases. However, the outcomes for patients after a lung transplant procedure are poor when compared with those after transplant of other solid organs. Primary graft dysfunction (PGD), also known as ischemia–reperfusion injury, pulmonary reimplantation response or early allograft failure, remains the most common early complication of lung transplanFrom the Department of Cardiovascular Surgery, University of Minnesota, Minneapolis, Minnesota. Submitted May 17, 2005; revised August 9, 2005; accepted November 2, 2005. Reprint requests: Peter S. Dahlberg, PhD, Department of Cardiothoracic Surgery, University of Minnesota, 420 Delaware Street SE, MMC 207, Minneapolis, MN 55435. Tel: 612-625-5636. Fax: 612-6259657. E-mail: [email protected] Copyright © 2006 by the International Society for Heart and Lung Transplantation. 1053-2498/06/$–see front matter. doi:10.1016/ j.healun.2005.11.436
tation and contributes significantly to mortality after the procedure.1,2 PGD occurs in 10% to 30% of patients and is marked by hypoxemia, increased capillary permeability, ventilation–perfusion mismatch and radiographic infiltrates, all occurring within the first 72 hours posttransplant. Clinically, patients with PGD have a spectrum of acute disease. Most have transient hypoxemia with mild infiltrates on chest X-ray, whereas others may develop life-threatening acute respiratory distress syndrome (ARDS) requiring aggressive intervention, including extracorporeal membrane oxygenation (ECMO) support.3 PGD, however, is recognized as a potentially reversible condition. Therefore, risk factor identification coupled with timely intervention, guided by reliable measures of allograft dysfunction, should improve surgical outcomes. Various studies have characterized the impact of PGD on short- and long-term patient outcomes.4 –9 Interpretation of these studies has, however, been hampered by 371
372
Prekker et al.
The Journal of Heart and Lung Transplantation April 2006
Table 1. Grading of Primary Graft Dysfunction (PGD) After Lung Transplantation
Grade 0 Grade 1 Grade 2 Grade 3 Time-points Worst grade T(0–48)
P/F P/F P/F P/F T0, N/A
ISHLT PGD definition ratio ⬎300 CXR clear ratio ⬎300 CXR infiltrate ratio 200–299 ratio ⬍200 T12, T24, T48, T72
P/F score (modified ISHLT) PGD definition N/A P/F ratio ⬎300 P/F ratio 200–299 P/F ratio ⬍200 T0, T12, T24, T48 Highest P/F grade between T0 and T48
Oxygenation index PGD definition N/A OI ⱕ2.4 OI 2.5–3.9 OI ⱖ4.0 T0, T12, T24, T48 Highest OI grade between T0 and T48
Grading criteria and the time-points for grade assignment are shown for the ISHLT PGD grading system, our modified ISHLT grading system, and an alternative to the P/F ratio– based systems, an oxygenation index score. P/F, partial pressure of arterial oxygen to fraction of inspired oxygen ratio; CXR, chest X-ray. OI, oxygenation index.
the use of variable definitions of the syndrome and non-compatible measures of PGD severity. Surrogate measurements for quantifying PGD include the partial pressure of arterial oxygen to fraction of inspired oxygen (P/F) ratio, the oxygenation index, the alveolar to arterial oxygenation (A-a) gradient and lung compliance. These surrogates have been incorporated into scoring systems for PGD.10,11 The scoring systems have been shown to predict short-term outcome in lung transplant patients. However, a consensus among transplant centers regarding the most useful assessment of the severity of PGD, and the optimal timing for such an assessment, is lacking. To address this problem, a working group from the International Society for Heart and Lung Transplantation (ISHLT) recently proposed standards for defining PGD that allow the clinician to calculate a PGD severity score between 0 and 3 at pre-determined time-points in the post-transplant period (Table 1).12 The purposes of this study were to: (1) validate the utility of the modified International Society for Heart and Lung Transplantation (ISHLT) grading system for PGD, the P/F grade, with respect to short-term outcomes; (2) compare the performance of the P/F grading system to other measures of early allograft function; and (3) assess the long-term survival and pulmonary function of patients who develop the most severe grade of PGD within the first 48 hours after transplantation. METHODS Patients The institutional review board at the University of Minnesota approved the study. All patients who underwent single or bilateral single-lung transplantation between January 1, 1992 and December 31, 2004 were eligible for our study. We performed a total of 402 single and bilateral single-lung transplants using lungs from 318 donors. There were 8 patients who underwent re-transplantation. Pre-transplant evaluations and hospital records were reviewed, and pulmonary function test results were collected for all patients at 1 and 2 years post-transplantation. The donor characteristics
examined were demographics, cause of death, cytomegalovirus (CMV) exposure status, days on ventilator before retrieval of the organs, bronchoscopic findings, Gram stain and culture results, blood gas analysis and graft ischemic times. Recipient characteristics examined were demographics, pre-operative hemodynamics, co-morbidities, CMV status, pulmonary function tests, whether cardiopulmonary bypass was used for the operation, and the preservation solution used for the organs. The lung transplantation protocol at our institution, including details of organ retrieval, operative technique and post-operative immunosuppression, has been previously reported.13 Of note, transplanted lungs were preserved with modified Euro-Collins solution for the initial 329 patients in our series; we switched to a low-potassium dextran solution (Perfadex) in July 2002 for the remaining 73 transplants. ISHLT Grading System and PGD Definition Patients were assigned PGD grades on the basis of guidelines issued by the 2004 ISHLT working group conference on the subject.12 Patients who were extubated at a specific time-point were also classified as having Grade 1 PGD. Patients receiving ECMO (23 of 402, or 6% of patients) were automatically classified as having Grade 3 PGD while receiving support. We made two changes to the ISHLT grading system for purposes of our analysis (Table 1). First, the original grading system proposed including five time-points up to 72 hours post-transplantation for grading PGD severity. Most patients in our series did not have arterial blood gases measured at T72, or 72 hours post-transplant. In fact, only 127 patients (32%) remained intubated in the ICU at this time-point. Therefore, the time-points chosen for analysis were at ICU arrival (T0) and 12 hours (T12), 24 hours (T24) and 48 hours (T48) post-transplantation. The worst grade T(0 – 48) reflects the most severe grade recorded during the 48-hour period. The second modification we made to the scoring system was to classify any patient with a P/F ⬎300 as having
The Journal of Heart and Lung Transplantation Volume 25, Number 4
Prekker et al.
373
Table 2. Characteristics of the Study Patients
Male Age (y) COPD IPF CF PPH PA, pre-operative (mm Hg) PA, OR (mm Hg) CPB Ischemic time (min) ICU days Hospital days
All patients (n ⫽ 402) 46% 50 ⫾ 11 61% 12% 11% 6% 25 ⫾ 11 27 ⫾ 11 30% 287 ⫾ 92 3 (IQR, 2-8) 14 (IQR 10-21)
SLT (n ⫽ 256) 45% 55 ⫾ 7b 76%a 16%a ⬍1%a ⬍1%a 23 ⫾ 7b 25 ⫾ 7b 12%a 245 ⫾ 62b 3 (IQR 2-6) 12 (IQR, 10-19)
BLT (n ⫽ 146) 47% 41 ⫾ 10 37% 4% 30% 16% 29 ⫾ 15 30 ⫾ 16 60% 357 ⫾ 90 5 (IQR 3-11) 16 (IQR 12-24)
Data are shown as percentage, mean ⫾ SD, or median (IQR). SLT, single lung transplant; BLT, bilateral single lung transplant; COPD, chronic obstructive pulmonary disease; IPF, idiopathic pulmonary fibrosis; CF, cystic fibrosis; PPH, primary pulmonary hypertension; PA, pulmonary arterial pressure; OR, operating room; CPB, cardiopulmonary bypass; IQR, interquartile range. a p ⬍ 0.05 vs BLT by chi-square test. b p ⬍ 0.05 vs BLT by Student’s t-test.
Grade 1 PGD (Table 1). We did not analyze chest X-rays for the presence or absence of infiltrates. Oxygenation Index ISHLT guidelines recommend using P/F ratios for grading PGD severity; we also examined the predictive value of a system based on oxygenation index (OI), which incorporates the P/F ratio as a measure of gas exchange and the mean airway pressure as a measure of pulmonary mechanics [OI ⫽ mean airway pressure/ (P/F ratio)]. We calculated the patient’s OI at the four time-points after transplantation, and an OI grade was assigned at each point (Table 1). In addition, the worst grade was recorded, which reflects the highest OI grade achieved within 48 hours of transplantation. Grade 1 to 3 cut-offs were assigned such that each composite grade had an approximately equal number of patients. Statistics Donor- and patient-related characteristics (including some measure of PGD), were analyzed separately for their association with 90-day mortality and long-term survival. These bivariate associations were tested for statistical significance using either Pearson’s chi-square test, for nominal and ordinal data, or the independentsample t-test for continuous variables. Analyses were performed using SAS for Windows, version 9.1 (SAS Institute, Inc., Cary, NC). Multivariate logistic regression was used for analysis of 90-day mortality. Final results were expressed as multivariate odds ratio (OR) and 95% confidence interval (CI). The OR can be interpreted as the increased or decreased odds of 90-day mortality for those patients with the characteristic compared with the odds of the
event for those without the characteristic (the reference category). The c-statistic was used to measure the discrimination of the final model. Values approaching ⱖ0.70 were considered the minimum necessary for a sound model. The Hosmer–Lemeshow test for goodness of fit was used to measure the calibration of the final model. A statistically non-significant chi-square is accepted for adequate agreement between the observed and expected events. Time-to-event analysis was used to analyze survival beyond 90 days. Risk factors were screened for inclusion using the Kaplan–Meier actuarial method and the log-rank test. Cox proportional hazards regression was used for the multivariate analysis. The final model was tested for the proportionality assumption. Final results were expressed as relative risk (RR) with the 95% confidence interval. RESULTS Patient Characteristics Table 2 lists the characteristics of our overall patient cohort, as well as data grouped by single-lung or bilateral single-lung transplantation. During the 13-year study period, we performed 402 lung transplant procedures from cadaveric donors; 256 patients (64%) received single-lung transplants (SLTs) and 146 (36%) received bilateral single-lung transplants (BLTs). Eight patients (7 SLTs and 1 BLT) underwent re-transplantation (2%). The mean recipient age at the time of transplantation was 50 ⫾ 11 years (range 7 to 68 years), and 46% were men. The indication to undergo transplantation included chronic obstructive pulmonary disease (COPD; 61%), idiopathic pulmonary fibrosis (IPF; 12%), cystic fibrosis (CF; 11%), primary pulmonary hypertension (PPH; 6%) and miscellaneous
374
Prekker et al.
disorders (7%). Patients diagnosed with COPD and IPF underwent SLT significantly more often than BLT, whereas BLT was favored over SLT for other causes of end-stage lung disease (Table 2). Mean allograft ischemic times were 245 ⫾ 62 minutes (range 120 to 428 minutes) for SLTs and 357 ⫾ 90 minutes (range 194 to 697 minutes) for the second lung transplanted in the BLT pair. Pre-transplant mean pulmonary artery (PA) pressures, as measured by right-heart catheterization performed as part of a pre-transplant evaluation, were significantly lower in SLT patients (23 ⫾ 7 mm Hg, range 9 to 58 mm Hg) than BLT patients (29 ⫾ 15 mm Hg, range 7 to 81 mm Hg) (p ⬍ 0.0001). Mean PA pressures tended to increase in the interval between pre-transplant catheterization and PA pressure measurement in the operating room by Swan–Ganz catheter, but the difference was not statistically significant (Table 2). Cardiopulmonary bypass (CPB) was the only other significant factor that occurred more frequently in patients having BLT (60%) than SLT (12%) (p ⬍ 0.0001).
Figure 1. Spectrum of post-operative oxygenation in 402 lung transplant patients at ICU arrival (T0). Arterial blood gas samples were collected at ICU arrival and PaO2/FIO2 (P/F) ratios and oxygenation indices were calculated. All patients on ECMO support were excluded. (a) The spectrum of P/F ratios is shown grouped by the modified ISHLT primary graft dysfunction (PGD) grading system (shading). (b) Oxygenation index values are shown grouped by an alternate grading system (shading). *Oxygenation index values of ⬎6.0 at T0 (n ⫽ 68, range 6.2 to 35.4) are not shown.
The Journal of Heart and Lung Transplantation April 2006
Figure 2. Prevalence of P/F Grades 1 to 3 at four time-points post-transplantation and the worst score T(0 – 48). T0 represents the first measurement after ICU arrival; T12, T24 and T48 are 12, 24 and 48 hours after transplantation, respectively. T(0 – 48) represents a worst score, the highest modified ISHLT grade assigned between T0 and T48.
Prevalence of PGD We recorded the P/F ratio and calculated OI values at the four study time-points (T0, T12, T24 and T48) for all patients. The distribution of values at the time of ICU arrival is shown in Figure 1. Figure 2 shows the prevalence of PGD grades at the four time-points posttransplantation and for the worst score T(0 – 48). There were 162 patients (44%) extubated by T24, and 248 patients (67%) by T48. The majority of patients at each time-point had Grade 1 PGD, representing a P/F ratio ⬎300 or freedom from mechanical ventilation, and the prevalence of Grade 1 PGD increased over time. The prevalence of severe PGD (Grade 3) decreased from T0 (25%) to T48 (15%). Grouping patients by their worst score T(0 – 48), the highest ISHLT grade assigned within 48 hours, yielded evenly distributed groups: 152 patients with Grade 1 (38%); 109 with Grade 2 (28%); and 133 with Grade 3 (34%) (Figure 2). Similar numbers of SLT (32%) and BLT (37%) patients had a composite score of Grade 3. Grading the severity of PGD using the oxygenation index revealed similar trends in PGD prevalence. The percentage of patients with severe PGD (OI ⱖ4) also decreased over time, from 34% of patients at T0 to 16% at T48. We grouped patients by their worst OI grade in the 48 hours after transplant. These groups included 85 patients with OI Grade 1 (22%), 127 with OI Grade 2 (32%) and 180 with OI Grade 3 (46%). Mortality at 90 Days Within 90 days after surgery, 49 of the 402 patients (12%) died. After SLT, 28 of 256 died (11%), and 21 of 146 (14%) died after BLT (p ⫽ 0.31, chi-square test). Table 3 shows 90-day mortality rates grouped by PGD grade at each of the four time-points and by worst score T(0 – 48), using both the P/F ratio (modified ISHLT) and
The Journal of Heart and Lung Transplantation Volume 25, Number 4
Prekker et al.
Table 3. Ninety-day Mortality and Severity of Lung Dysfunction P/F grade Grade 1 Grade 2 Grade 3 OI grade Grade 1 Grade 2 Grade 3
T0
T24
T48
T(0–48)
10% 11% 18% p ⫽ 0.12
8% 7% 28% p ⬍ 0.0001
7% 12% 33% p ⬍ 0.0001
10% 8% 17% p ⫽ 0.05
9% 11% 17% p ⫽ 0.11
7% 14% 18% p ⫽ 0.018
7% 15% 25% p ⫽ 0.0002
5% 11% 16% p ⫽ 0.025
The log-rank p values compare differences in 90-day mortality rates using Kaplan-Meier analysis by level of graft dysfunction. OI, oxygenation index.
the OI grading criteria. At the T24 and T48 time-points, we observed a significantly higher 90-day mortality rate in the group of patients with P/F Grade 3 as compared to those with Grades 1 and 2 (Table 3). Although the trend in 90-day death for patients grouped by PGD grade at ICU arrival (T0) was similar to that seen at the other two time-points, it did not reach statistical significance. In the sub-group of patients undergoing SLT, the same pattern was observed. SLT patients with Grade 3 PGD at the T24 and T48 time-points had a significantly higher 90-day death rate compared to patients without Grade 3 PGD, whereas the T0 trends did not reach the level of statistical significance (8.8% for no Grade 3 vs 18% for Grade 3: p ⫽ 0.068, chi-square test). In the BLT sub-group, development of Grade 3 PGD at T24 and T48 was a significant predictor of 90-day death. Employing OI grading criteria, the greatest differentiation in 90-day mortality rate between patients with mild, moderate and severe PGD occurred at T48. The OI worst score T(0 – 48) was able to stratify patients into grades with significantly different 90-day death rates, and in this sense was superior to the P/F worst score T(0 – 48) (Table 3). Long-term Survival Long-term survival curves are shown in Figure 3. The 5and 10-year survival rates for the overall group were 55% and 24%, respectively (Figure 3a). The median survival was 5.6 years (95% CI 4.9 to 6.6). Figure 3b shows survival curves for all patients grouped by their worst score T(0 – 48). We did not observe any difference in survival between Grades 1 and 2; however, Grade 3 patients had reduced longterm survival when compared with the combined Group 1 and Group 2 patients (Grade 3 median survival 7.0 years [95% CI 5.6 to 7.8] vs 5.2 years for no Grade 3 [2.9 to 4.7], p ⱕ 0.0001, log rank test). This difference was due in part to an increased risk of peri-operative death; however, the curves for 90-day survivors with
375
and without P/F Grade 3 (Figure 3c) highlight the validity of the Grade 3 PGD score as a predictor of poorer long-term survival. An earlier identifier of poor outcome than the worst score T(0 – 48) would be useful to guide timely clinical intervention for patients at risk. We analyzed the predictive value of the earliest time-point, T0. The survival curve of the cohort grouped by their PGD grade at the earliest time-point, T0, is shown in Figure 3d. There was no difference in survival between Grades 1 and 2; however, the T0 Grade 3 group had significantly poorer long-term survival compared with the combined Group 1 and Group 2 patients (median survival 6.5 years [95% CI 5.4 to 7.6] for no Grade 3, 3.8 years [2.2 to 4.9] for Grade 3, p ⫽ 0.0006, log rank test). P/F Grades 1 and 2 did not stratify patients into groups with differential long-term mortality at T0 or when grouped by worst score T(0 – 48). In addition, the predictive power of Grades 1 and 2 did not improve at T24 or T48 (data not shown). Therefore, we examined OI grade at T0 and as a worst score T(0 – 48) as a predictor of long-term survival. Patients classified as OI Grade 2 at T0 appeared to have an intermediate survival between Grade 1 and Grade 3; however, the differences in survival curves did not reach statistical significance and may have been due to chance (Figure 3e). Multivariate Analysis We analyzed T(0 – 48) Grade 3, T0 Grade 3 and T0 OI as predictors of outcome in multivariate models with other potential donor and recipient risk factors. The hazards of variables, such as need for cardiopulmonary bypass during the transplant operation or the development of Grade 3 graft dysfunction, are not necessarily constant over time (being greater in the peri-operative period); therefore, early outcome was modeled by logistic regression using 90-day death as the end-point, and late survival by a Cox proportional hazards analysis of peri-operative survivors. Only clinically meaningful variables associated with the respective end-points in a bivariate analysis at the p ⫽ 0.1 level were included in the models. Results of the analysis are shown in Table 4. Secondary Outcome Measures We evaluated the PGD grading system as a predictor of clinical outcomes apart from overall mortality, including patients’ length of ICU and hospital stay, as shown in Table 5. Patients who developed Grade 3 PGD by P/F score within 48 hours of surgery had significantly longer ICU and hospital stays vs patients classified as Grades 1 or 2. This was true regardless of whether SLT or BLT was performed. Of the 131 patients who continued to require mechanical ventilation at T48, 83 patients (63%) had Grade 3 PGD as their worst grade T(0 – 48), as compared with 35 patients (27%) with
376
Prekker et al.
The Journal of Heart and Lung Transplantation April 2006
Figure 3. (a) Long-term survival of the overall group. The numbers of patients at time of transplant and at the 5-year and 10-year time-points are listed at the bottom. The 5-year actuarial survival rate was 55%, and at 10 years it was 24%. (b) Overall survival of the 394 patients with PaO2/FIO2 data available stratified by worst P/F score T(0 – 48) reflecting the highest modified ISHLT grade assigned between T0 and T48. The numbers of patients at the time of transplant and at the 5-year and 10-year time-points are listed at the bottom. (c) Overall survival of the 348 patients who survived the 90-day peri-operative period, grouped by worst P/F score T(0 – 48). (d) Overall survival of the 394 patients grouped by P/F grade at T0. (e) Overall survival of the 389 patients with oxygenation index data available, grouped by T0 oxygenation index grade. OI, oxygenation index.
Grade 2 PGD, and only 13 patients (9%) were classified as Grade 1 PGD. Patients in the study who survived the peri-operative period were followed with pulmonary function testing at 1 year and 2 years after transplant. Means for forced expiratory volume in 1 second (FEV1) for SLT and BLT
patients are listed in Table 5, grouped by the worst P/F grade T(0 – 48). SLT patients had significantly poorer pulmonary function at both 1 and 2 years compared with patients who underwent BLT (Table 5). Increasing severity of PGD within the first 48 hours after lung transplantation was associated with decreasing pulmo-
The Journal of Heart and Lung Transplantation Volume 25, Number 4
Prekker et al.
Table 4. Multivariate Analysis No T(0–48) Grade 3 PGD BLT No CPB High pre-operative PA (mean) Recipient age Donor age Recipient male
p-value 0.03 ⬍0.01 ⬍0.01 0.04 NS NS NS
RR 0.85 0.61 0.67 1.02
95% CI 0.73–0.98 0.47–0.78 0.55–0.81 1.01–1.04
Significant univariate predictors of increased 90-day and long-term mortality are shown. Those variables that reached significance in our multivariate model are accompanied by p-values as well as relative risk and 95% confidence intervals. RR, relative risk; CI, confidence interval; SLT, single-lung transplant; CPB, cardiopulmonary bypass; PA, pulmonary arterial pressure.
nary function at both 1 and 2 years. However, differences among mean FEV1 values between those with and without severe PGD only reached statistical significance among BLT patients at 1 year (p ⫽ 0.002). DISCUSSION A useful system for grading PGD after lung transplantation must have several features. First, it should accurately stratify patients into groups with different perioperative death rates and perhaps different long-term survival rates. Second, it should rapidly identify patients who are developing the most severe forms of PGD, who might benefit from an effective intervention. Third, the measurement used to calculate the grade should be objective, reliably obtained, and simple to record, report and track within institutional databases and larger registries. The P/F grading system described in our work satisfies these requirements. The power of the PGD grading system (modified ISHLT criteria) to predict differential mortality is greatest when those patients with severe PGD (Grade 3, P/F ⬍200) are compared to those without severe PGD. Furthermore, Grade 3 PGD is an independent risk factor for poor long-term survival, when analyzed as the worst score T(0 – 48) in our multivariate model. Other inde-
377
pendent risk factors in our multivariate analysis were SLT, use of CPB, and increased pre-operative mean pulmonary artery pressure. The different components of the PGD definition are useful for different purposes. The T48 (and presumably T72) time-points are probably the end-points that best capture patients with the most significant forms of PGD. In our series, T48 demonstrated the clearest differentiation of 90-day death rates between mild PGD (Grade 1, 7%), moderate (Grade 2, 12%) and severe graft dysfunction (Grade 3, 33%) (p ⬍ 0.0001). The T0 P/F measurement is the best way to rapidly identify patients who might be destined to have a poor outcome and who might benefit from intervention. Additional early trend analysis might more accurately stratify this group of patients with respect to short-term mortality because one would expect that a patient with a rapidly improving ratio would have a more favorable prognosis than the overall group. Conversely, a patient with an unfavorable 12-hour change in their P/F ratio would be expected to have a higher operative mortality rate. The T0 data set is also useful in the sense that it is the most complete because nearly all transplant patients are mechanically ventilated when they arrive in the ICU. Grouping extubated patients in the favorable category at later time-points might bias the otherwise objective measurement of the P/F ratio. There was no obvious advantage to grouping patients by OI score, which incorporates airway pressures in addition to gas exchange into its definition.14 The qualitative stratification of short- and medium-term outcomes of Grade 1 and Grade 2 patients was improved as compared with the P/F grade, but the differences in curves did not reach the level of statistical significance. Our study has several limitations, the most obvious of which is that we did not strictly adhere to the ISHLT working group criteria for the assignment of grades. We did not collect data out to the 72-hour time-point nor did we analyze chest X-rays for the presence or absence
Table 5. Secondary Outcomes: ICU and Hospital Length of Stay, Pulmonary Function Tests ICU days Hospital days 1-year FEV1 SLT (n ⫽ 163) BLT (n ⫽ 103) 2-year FEV1 SLT (n ⫽ 130) BLT (n ⫽ 87)
Grade 1 4.4 ⫾ 1.0 14 ⫾ 1.5
Grade 2 5.4 ⫾ 1.1 16 ⫾ 1.7
Grade 3 12 ⫾ 1.0a 25 ⫾ 1.5a
p ⬍ 0.0001 p ⬍ 0.0001
58 ⫾ 1.7 92 ⫾ 4.3
52 ⫾ 1.9 87 ⫾ 4.4
52 ⫾ 1.8 71 ⫾ 4.5b
p ⫽ 0.056 p ⫽ 0.002
56 ⫾ 2.0 93 ⫾ 5.2
52 ⫾ 2.3 85 ⫾ 5.6
51 ⫾ 2.1 75 ⫾ 5.4
NS p ⫽ 0.059
Secondary outcomes are shown grouped by worst P/F grade T(0 – 48). ICU and hospital length of stay are listed as mean number of days. Patients were followed routinely with pulmonary function testing. Values at 1 and 2 years after transplant are reported as the percent of the predicted value based on the height and weight of the patient. FEV1, forced expiratory volume in 1 second; SLT, single-lung transplant; BLT, bilateral single-lung transplant. a p ⬍ 0.0001 vs Grades 1 and 2 PGD. b p ⫽ 0.002 vs Grades 1 and 2 PGD.
378
Prekker et al.
of infiltrates. We would expect that the presence of Grade 3 PGD at the 72-hour measurement would be predictive of an even greater peri-operative mortality rate than the 48-hour measurement. The T72 value might also correlate more with longer-term survival and with pulmonary function than the values that we analyzed. With regard to the chest X-rays, we found that most lung transplant patients do not have entirely clear films in the first 24 hours after transplant. With our institution’s change to electronic imaging, almost all of the ⬎2000 X-rays that would require analysis are either no longer available or are stored in warehouse archives. A clear chest X-ray 48 or 72 hours after a lung transplant might occur more frequently than it does in the earlier post-operative period and could enhance the predictive value of the ISHLT scoring system. However, this piece of recorded information is more subjective than the P/F ratio and would probably not be reliably reported to the registries by all centers. In conclusion, the development of severe (ISHLT Grade 3) primary graft dysfunction is associated with high operative mortality and decreased long-term survival. In the sub-set of patients who undergo bilateral single-lung transplantation, Grade 3 PGD is also associated with reduced pulmonary function at 1 and 2 years. In a multivariate analysis, risk factors for death in addition to Grade 3 PGD were the need for CPB during the transplant, having undergone single-lung transplant, and increasing pre-operative pulmonary artery pressure. Our results support the vision of the ISHLT working group to use the grading system as a standard tool in clinical care and in clinical trials to improve the outlook for patients undergoing pulmonary transplantation. Early identification of patients with the most severe forms of PGD may allow for the use of selective management strategies such as altered ventilation to reduce airway pressure as in ARDS, beta-blockade to limit perfusion of already injured lungs, pharmacologic agents to reduce ischemia–reperfusion injury,15–17 or even extracorporeal membrane oxygenation to allow the lungs a period of low-perfusion recovery for the most severe forms of PGD.3,18 REFERENCES 1. Trulock EP, Edwards LB, Taylor DO, et al. The registry of the International Society for Heart and Lung Transplantation: twenty-first official adult lung and heart–lung transplant report—2004. J Heart Lung Transplant 2004;23: 804 –15. 2. Christie JD, Kotloff RM, Ahya VN, et al. The effect of primary graft dysfunction on survival after lung transplantation. Am J Respir Crit Care Med 2005;171:1312– 6.
The Journal of Heart and Lung Transplantation April 2006
3. Dahlberg PS, Prekker ME, Herrington CS, Hertz MI, Park SJ. Medium-term results of extracorporeal membrane oxygenation for severe acute lung injury after lung transplantation. J Heart Lung Transplant 2004;23:979 – 84. 4. Christie JD, Bavaria JE, Palevsky HI, et al. Primary graft failure following lung transplantation. Chest 1998;114: 51– 60. 5. Christie JD, Kotloff RM, Pochettino A, et al. Clinical risk factors for primary graft failure following lung transplantation. Chest 2003;124:1232– 41. 6. Christie JD, Sager JS, Kimmel SE, et al. Impact of primary graft failure on outcomes following lung transplantation. Chest 2005;127:161–5. 7. Fiser SM, Kron IL, Long SM, Kaza AK, Kern JA, Tribble CG. Early intervention after severe oxygenation index elevation improves survival following lung transplantation. J Heart Lung Transplant 2001;20:631– 6. 8. Khan SU, Salloum J, O’Donovan PB, et al. Acute pulmonary edema after lung transplantation: the pulmonary reimplantation response. Chest 1999;116:187–94. 9. King RC, Binns OA, Rodriguez F, et al. Reperfusion injury significantly impacts clinical outcome after pulmonary transplantation. Ann Thorac Surg 2000;69:1681–5. 10. Thabut G, Vinatier I, Stern JB, et al. Primary graft failure following lung transplantation: predictive factors of mortality. Chest 2002;121:1876 – 82. 11. 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. 12. Christie JD, Carby M, Bag R, Corris P, Hertz M, Weill D. 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 –9. 13. Dahlberg PS, Prekker ME, Hertz MI, Thompson DJ, Park SJ. Recent trends in lung transplantation: the University of Minnesota experience. Clin Transplants 2002;XX:243–51. 14. Green TP, Timmons OD, Fackler JC, et al. The impact of extracorporeal membrane oxygenation on survival in pediatric patients with acute respiratory failure. Pediatric Critical Care Study Group. Crit Care Med 1996;24:323–9. 15. de Perrot M, Liu M, Waddell TK, Keshavjee S. Ischemia– reperfusion-induced lung injury. Am J Respir Crit Care Med 2003;167:490 –511. 16. Meade MO, Granton JT, Matte-Martyn A, et al. A randomized trial of inhaled nitric oxide to prevent ischemia– reperfusion injury after lung transplantation. Am J Respir Crit Care Med 2003;167:1483– 89. 17. Keshavjee S, Davis RD, Zamora MR, de Perrot M, Patterson GA. A randomized, placebo-controlled trial of complement inhibition in ischemia–reperfusion injury after lung transplantation in human beings. J Thorac Cardiovasc Surg 2005;129:423– 8. 18. Oto T, Rosenfeldt F, Rowland M, et al. Extracorporeal membrane oxygenation after lung transplantation: evolving technique improves outcomes. Ann Thorac Surg 2004; 78:1230 –5.
Validation of the Proposed International Society for Heart and Lung Transplantation Grading System for Primary Graft Dysfunction After Lung Transplantation Matthew E. Prekker, MD, D. S. Nath, MD, A. R. Walker, MD, A. C. Johnson, MD, M. I. Hertz, MD, C. S. Herrington, MD, D. M. Radosevich, PhD, and Peter S. Dahlberg, MD, PhD Background: A scoring system was recently proposed to grade the severity of primary graft dysfunction (PGD), a frequent early complication of lung transplantation. The purposes of this study are to: (1) validate the PGD grading system with respect to patient outcomes; and (2) compare the performance of criteria employing the arterial oxygenation to fraction of inspired oxygen (P/F) ratio to an alternative grading system employing the oxygenation index (OI). Methods: We retrospectively reviewed the medical records of 402 patients having undergone lung transplantation at our institution from 1992 through 2004. The ISHLT PGD grading system was modified and grades were assigned up to 48 hours post-transplantation as follows: Grade 1 PGD, P/F ⬎300; Grade 2, P/F 200 to 300; and Grade 3, P/F ⬍200. A worst score T(0 – 48) was also assigned, which reflects the highest grade recorded between T0 and T48. Results: The prevalence of severe PGD (P/F Grade 3) declined after transplant, from 25% at T0 to 15% at T48. Grouping patients by P/F grade at T48 demonstrated the clearest differentiation of 90-day death rates (Grade 1, 7%; Grade 2, 12%; Grade 3, 33%) (p ⫽ 0.0001). T48 OI grade also differentiates 90-day death rates. There was no difference in longer-term survival between patients with PGD Grades 1 and 2. OI grade at T0 qualitatively improved differential mortality between Grades 1 and 2; however, the differences did not reach statistical significance. Patients with a worst score T(0 – 48) of Grade 3 PGD did have significantly decreased long-term survival, as well as longer ICU and hospital stay, when compared with Grades 1 and 2 PGD. Significant risk factors for short- and long-term mortality in our multivariate model were P/F Grade 3 [worst score T(0 – 48) as well as T0 grade], single-lung transplant, use of cardiopulmonary bypass and high pre-operative mean pulmonary artery pressure. Conclusions: There is an increased risk of short- and long-term mortality and length of hospital stay associated with severe (Grade 3) PGD. The proposed ISHLT grading system can rapidly identify patients with poor outcomes who may benefit from early, aggressive treatment. Refinement of the scoring system may further improve patient risk stratification. J Heart Lung Transplant 2006;25:371– 8. Copyright © 2006 by the International Society for Heart and Lung Transplantation.
Transplantation is the only effective treatment for several end-stage lung diseases. However, the outcomes for patients after a lung transplant procedure are poor when compared with those after transplant of other solid organs. Primary graft dysfunction (PGD), also known as ischemia–reperfusion injury, pulmonary reimplantation response or early allograft failure, remains the most common early complication of lung transplanFrom the Department of Cardiovascular Surgery, University of Minnesota, Minneapolis, Minnesota. Submitted May 17, 2005; revised August 9, 2005; accepted November 2, 2005. Reprint requests: Peter S. Dahlberg, PhD, Department of Cardiothoracic Surgery, University of Minnesota, 420 Delaware Street SE, MMC 207, Minneapolis, MN 55435. Tel: 612-625-5636. Fax: 612-6259657. E-mail: [email protected] Copyright © 2006 by the International Society for Heart and Lung Transplantation. 1053-2498/06/$–see front matter. doi:10.1016/ j.healun.2005.11.436
tation and contributes significantly to mortality after the procedure.1,2 PGD occurs in 10% to 30% of patients and is marked by hypoxemia, increased capillary permeability, ventilation–perfusion mismatch and radiographic infiltrates, all occurring within the first 72 hours posttransplant. Clinically, patients with PGD have a spectrum of acute disease. Most have transient hypoxemia with mild infiltrates on chest X-ray, whereas others may develop life-threatening acute respiratory distress syndrome (ARDS) requiring aggressive intervention, including extracorporeal membrane oxygenation (ECMO) support.3 PGD, however, is recognized as a potentially reversible condition. Therefore, risk factor identification coupled with timely intervention, guided by reliable measures of allograft dysfunction, should improve surgical outcomes. Various studies have characterized the impact of PGD on short- and long-term patient outcomes.4 –9 Interpretation of these studies has, however, been hampered by 371
372
Prekker et al.
The Journal of Heart and Lung Transplantation April 2006
Table 1. Grading of Primary Graft Dysfunction (PGD) After Lung Transplantation
Grade 0 Grade 1 Grade 2 Grade 3 Time-points Worst grade T(0–48)
P/F P/F P/F P/F T0, N/A
ISHLT PGD definition ratio ⬎300 CXR clear ratio ⬎300 CXR infiltrate ratio 200–299 ratio ⬍200 T12, T24, T48, T72
P/F score (modified ISHLT) PGD definition N/A P/F ratio ⬎300 P/F ratio 200–299 P/F ratio ⬍200 T0, T12, T24, T48 Highest P/F grade between T0 and T48
Oxygenation index PGD definition N/A OI ⱕ2.4 OI 2.5–3.9 OI ⱖ4.0 T0, T12, T24, T48 Highest OI grade between T0 and T48
Grading criteria and the time-points for grade assignment are shown for the ISHLT PGD grading system, our modified ISHLT grading system, and an alternative to the P/F ratio– based systems, an oxygenation index score. P/F, partial pressure of arterial oxygen to fraction of inspired oxygen ratio; CXR, chest X-ray. OI, oxygenation index.
the use of variable definitions of the syndrome and non-compatible measures of PGD severity. Surrogate measurements for quantifying PGD include the partial pressure of arterial oxygen to fraction of inspired oxygen (P/F) ratio, the oxygenation index, the alveolar to arterial oxygenation (A-a) gradient and lung compliance. These surrogates have been incorporated into scoring systems for PGD.10,11 The scoring systems have been shown to predict short-term outcome in lung transplant patients. However, a consensus among transplant centers regarding the most useful assessment of the severity of PGD, and the optimal timing for such an assessment, is lacking. To address this problem, a working group from the International Society for Heart and Lung Transplantation (ISHLT) recently proposed standards for defining PGD that allow the clinician to calculate a PGD severity score between 0 and 3 at pre-determined time-points in the post-transplant period (Table 1).12 The purposes of this study were to: (1) validate the utility of the modified International Society for Heart and Lung Transplantation (ISHLT) grading system for PGD, the P/F grade, with respect to short-term outcomes; (2) compare the performance of the P/F grading system to other measures of early allograft function; and (3) assess the long-term survival and pulmonary function of patients who develop the most severe grade of PGD within the first 48 hours after transplantation. METHODS Patients The institutional review board at the University of Minnesota approved the study. All patients who underwent single or bilateral single-lung transplantation between January 1, 1992 and December 31, 2004 were eligible for our study. We performed a total of 402 single and bilateral single-lung transplants using lungs from 318 donors. There were 8 patients who underwent re-transplantation. Pre-transplant evaluations and hospital records were reviewed, and pulmonary function test results were collected for all patients at 1 and 2 years post-transplantation. The donor characteristics
examined were demographics, cause of death, cytomegalovirus (CMV) exposure status, days on ventilator before retrieval of the organs, bronchoscopic findings, Gram stain and culture results, blood gas analysis and graft ischemic times. Recipient characteristics examined were demographics, pre-operative hemodynamics, co-morbidities, CMV status, pulmonary function tests, whether cardiopulmonary bypass was used for the operation, and the preservation solution used for the organs. The lung transplantation protocol at our institution, including details of organ retrieval, operative technique and post-operative immunosuppression, has been previously reported.13 Of note, transplanted lungs were preserved with modified Euro-Collins solution for the initial 329 patients in our series; we switched to a low-potassium dextran solution (Perfadex) in July 2002 for the remaining 73 transplants. ISHLT Grading System and PGD Definition Patients were assigned PGD grades on the basis of guidelines issued by the 2004 ISHLT working group conference on the subject.12 Patients who were extubated at a specific time-point were also classified as having Grade 1 PGD. Patients receiving ECMO (23 of 402, or 6% of patients) were automatically classified as having Grade 3 PGD while receiving support. We made two changes to the ISHLT grading system for purposes of our analysis (Table 1). First, the original grading system proposed including five time-points up to 72 hours post-transplantation for grading PGD severity. Most patients in our series did not have arterial blood gases measured at T72, or 72 hours post-transplant. In fact, only 127 patients (32%) remained intubated in the ICU at this time-point. Therefore, the time-points chosen for analysis were at ICU arrival (T0) and 12 hours (T12), 24 hours (T24) and 48 hours (T48) post-transplantation. The worst grade T(0 – 48) reflects the most severe grade recorded during the 48-hour period. The second modification we made to the scoring system was to classify any patient with a P/F ⬎300 as having
The Journal of Heart and Lung Transplantation Volume 25, Number 4
Prekker et al.
373
Table 2. Characteristics of the Study Patients
Male Age (y) COPD IPF CF PPH PA, pre-operative (mm Hg) PA, OR (mm Hg) CPB Ischemic time (min) ICU days Hospital days
All patients (n ⫽ 402) 46% 50 ⫾ 11 61% 12% 11% 6% 25 ⫾ 11 27 ⫾ 11 30% 287 ⫾ 92 3 (IQR, 2-8) 14 (IQR 10-21)
SLT (n ⫽ 256) 45% 55 ⫾ 7b 76%a 16%a ⬍1%a ⬍1%a 23 ⫾ 7b 25 ⫾ 7b 12%a 245 ⫾ 62b 3 (IQR 2-6) 12 (IQR, 10-19)
BLT (n ⫽ 146) 47% 41 ⫾ 10 37% 4% 30% 16% 29 ⫾ 15 30 ⫾ 16 60% 357 ⫾ 90 5 (IQR 3-11) 16 (IQR 12-24)
Data are shown as percentage, mean ⫾ SD, or median (IQR). SLT, single lung transplant; BLT, bilateral single lung transplant; COPD, chronic obstructive pulmonary disease; IPF, idiopathic pulmonary fibrosis; CF, cystic fibrosis; PPH, primary pulmonary hypertension; PA, pulmonary arterial pressure; OR, operating room; CPB, cardiopulmonary bypass; IQR, interquartile range. a p ⬍ 0.05 vs BLT by chi-square test. b p ⬍ 0.05 vs BLT by Student’s t-test.
Grade 1 PGD (Table 1). We did not analyze chest X-rays for the presence or absence of infiltrates. Oxygenation Index ISHLT guidelines recommend using P/F ratios for grading PGD severity; we also examined the predictive value of a system based on oxygenation index (OI), which incorporates the P/F ratio as a measure of gas exchange and the mean airway pressure as a measure of pulmonary mechanics [OI ⫽ mean airway pressure/ (P/F ratio)]. We calculated the patient’s OI at the four time-points after transplantation, and an OI grade was assigned at each point (Table 1). In addition, the worst grade was recorded, which reflects the highest OI grade achieved within 48 hours of transplantation. Grade 1 to 3 cut-offs were assigned such that each composite grade had an approximately equal number of patients. Statistics Donor- and patient-related characteristics (including some measure of PGD), were analyzed separately for their association with 90-day mortality and long-term survival. These bivariate associations were tested for statistical significance using either Pearson’s chi-square test, for nominal and ordinal data, or the independentsample t-test for continuous variables. Analyses were performed using SAS for Windows, version 9.1 (SAS Institute, Inc., Cary, NC). Multivariate logistic regression was used for analysis of 90-day mortality. Final results were expressed as multivariate odds ratio (OR) and 95% confidence interval (CI). The OR can be interpreted as the increased or decreased odds of 90-day mortality for those patients with the characteristic compared with the odds of the
event for those without the characteristic (the reference category). The c-statistic was used to measure the discrimination of the final model. Values approaching ⱖ0.70 were considered the minimum necessary for a sound model. The Hosmer–Lemeshow test for goodness of fit was used to measure the calibration of the final model. A statistically non-significant chi-square is accepted for adequate agreement between the observed and expected events. Time-to-event analysis was used to analyze survival beyond 90 days. Risk factors were screened for inclusion using the Kaplan–Meier actuarial method and the log-rank test. Cox proportional hazards regression was used for the multivariate analysis. The final model was tested for the proportionality assumption. Final results were expressed as relative risk (RR) with the 95% confidence interval. RESULTS Patient Characteristics Table 2 lists the characteristics of our overall patient cohort, as well as data grouped by single-lung or bilateral single-lung transplantation. During the 13-year study period, we performed 402 lung transplant procedures from cadaveric donors; 256 patients (64%) received single-lung transplants (SLTs) and 146 (36%) received bilateral single-lung transplants (BLTs). Eight patients (7 SLTs and 1 BLT) underwent re-transplantation (2%). The mean recipient age at the time of transplantation was 50 ⫾ 11 years (range 7 to 68 years), and 46% were men. The indication to undergo transplantation included chronic obstructive pulmonary disease (COPD; 61%), idiopathic pulmonary fibrosis (IPF; 12%), cystic fibrosis (CF; 11%), primary pulmonary hypertension (PPH; 6%) and miscellaneous
374
Prekker et al.
disorders (7%). Patients diagnosed with COPD and IPF underwent SLT significantly more often than BLT, whereas BLT was favored over SLT for other causes of end-stage lung disease (Table 2). Mean allograft ischemic times were 245 ⫾ 62 minutes (range 120 to 428 minutes) for SLTs and 357 ⫾ 90 minutes (range 194 to 697 minutes) for the second lung transplanted in the BLT pair. Pre-transplant mean pulmonary artery (PA) pressures, as measured by right-heart catheterization performed as part of a pre-transplant evaluation, were significantly lower in SLT patients (23 ⫾ 7 mm Hg, range 9 to 58 mm Hg) than BLT patients (29 ⫾ 15 mm Hg, range 7 to 81 mm Hg) (p ⬍ 0.0001). Mean PA pressures tended to increase in the interval between pre-transplant catheterization and PA pressure measurement in the operating room by Swan–Ganz catheter, but the difference was not statistically significant (Table 2). Cardiopulmonary bypass (CPB) was the only other significant factor that occurred more frequently in patients having BLT (60%) than SLT (12%) (p ⬍ 0.0001).
Figure 1. Spectrum of post-operative oxygenation in 402 lung transplant patients at ICU arrival (T0). Arterial blood gas samples were collected at ICU arrival and PaO2/FIO2 (P/F) ratios and oxygenation indices were calculated. All patients on ECMO support were excluded. (a) The spectrum of P/F ratios is shown grouped by the modified ISHLT primary graft dysfunction (PGD) grading system (shading). (b) Oxygenation index values are shown grouped by an alternate grading system (shading). *Oxygenation index values of ⬎6.0 at T0 (n ⫽ 68, range 6.2 to 35.4) are not shown.
The Journal of Heart and Lung Transplantation April 2006
Figure 2. Prevalence of P/F Grades 1 to 3 at four time-points post-transplantation and the worst score T(0 – 48). T0 represents the first measurement after ICU arrival; T12, T24 and T48 are 12, 24 and 48 hours after transplantation, respectively. T(0 – 48) represents a worst score, the highest modified ISHLT grade assigned between T0 and T48.
Prevalence of PGD We recorded the P/F ratio and calculated OI values at the four study time-points (T0, T12, T24 and T48) for all patients. The distribution of values at the time of ICU arrival is shown in Figure 1. Figure 2 shows the prevalence of PGD grades at the four time-points posttransplantation and for the worst score T(0 – 48). There were 162 patients (44%) extubated by T24, and 248 patients (67%) by T48. The majority of patients at each time-point had Grade 1 PGD, representing a P/F ratio ⬎300 or freedom from mechanical ventilation, and the prevalence of Grade 1 PGD increased over time. The prevalence of severe PGD (Grade 3) decreased from T0 (25%) to T48 (15%). Grouping patients by their worst score T(0 – 48), the highest ISHLT grade assigned within 48 hours, yielded evenly distributed groups: 152 patients with Grade 1 (38%); 109 with Grade 2 (28%); and 133 with Grade 3 (34%) (Figure 2). Similar numbers of SLT (32%) and BLT (37%) patients had a composite score of Grade 3. Grading the severity of PGD using the oxygenation index revealed similar trends in PGD prevalence. The percentage of patients with severe PGD (OI ⱖ4) also decreased over time, from 34% of patients at T0 to 16% at T48. We grouped patients by their worst OI grade in the 48 hours after transplant. These groups included 85 patients with OI Grade 1 (22%), 127 with OI Grade 2 (32%) and 180 with OI Grade 3 (46%). Mortality at 90 Days Within 90 days after surgery, 49 of the 402 patients (12%) died. After SLT, 28 of 256 died (11%), and 21 of 146 (14%) died after BLT (p ⫽ 0.31, chi-square test). Table 3 shows 90-day mortality rates grouped by PGD grade at each of the four time-points and by worst score T(0 – 48), using both the P/F ratio (modified ISHLT) and
The Journal of Heart and Lung Transplantation Volume 25, Number 4
Prekker et al.
Table 3. Ninety-day Mortality and Severity of Lung Dysfunction P/F grade Grade 1 Grade 2 Grade 3 OI grade Grade 1 Grade 2 Grade 3
T0
T24
T48
T(0–48)
10% 11% 18% p ⫽ 0.12
8% 7% 28% p ⬍ 0.0001
7% 12% 33% p ⬍ 0.0001
10% 8% 17% p ⫽ 0.05
9% 11% 17% p ⫽ 0.11
7% 14% 18% p ⫽ 0.018
7% 15% 25% p ⫽ 0.0002
5% 11% 16% p ⫽ 0.025
The log-rank p values compare differences in 90-day mortality rates using Kaplan-Meier analysis by level of graft dysfunction. OI, oxygenation index.
the OI grading criteria. At the T24 and T48 time-points, we observed a significantly higher 90-day mortality rate in the group of patients with P/F Grade 3 as compared to those with Grades 1 and 2 (Table 3). Although the trend in 90-day death for patients grouped by PGD grade at ICU arrival (T0) was similar to that seen at the other two time-points, it did not reach statistical significance. In the sub-group of patients undergoing SLT, the same pattern was observed. SLT patients with Grade 3 PGD at the T24 and T48 time-points had a significantly higher 90-day death rate compared to patients without Grade 3 PGD, whereas the T0 trends did not reach the level of statistical significance (8.8% for no Grade 3 vs 18% for Grade 3: p ⫽ 0.068, chi-square test). In the BLT sub-group, development of Grade 3 PGD at T24 and T48 was a significant predictor of 90-day death. Employing OI grading criteria, the greatest differentiation in 90-day mortality rate between patients with mild, moderate and severe PGD occurred at T48. The OI worst score T(0 – 48) was able to stratify patients into grades with significantly different 90-day death rates, and in this sense was superior to the P/F worst score T(0 – 48) (Table 3). Long-term Survival Long-term survival curves are shown in Figure 3. The 5and 10-year survival rates for the overall group were 55% and 24%, respectively (Figure 3a). The median survival was 5.6 years (95% CI 4.9 to 6.6). Figure 3b shows survival curves for all patients grouped by their worst score T(0 – 48). We did not observe any difference in survival between Grades 1 and 2; however, Grade 3 patients had reduced longterm survival when compared with the combined Group 1 and Group 2 patients (Grade 3 median survival 7.0 years [95% CI 5.6 to 7.8] vs 5.2 years for no Grade 3 [2.9 to 4.7], p ⱕ 0.0001, log rank test). This difference was due in part to an increased risk of peri-operative death; however, the curves for 90-day survivors with
375
and without P/F Grade 3 (Figure 3c) highlight the validity of the Grade 3 PGD score as a predictor of poorer long-term survival. An earlier identifier of poor outcome than the worst score T(0 – 48) would be useful to guide timely clinical intervention for patients at risk. We analyzed the predictive value of the earliest time-point, T0. The survival curve of the cohort grouped by their PGD grade at the earliest time-point, T0, is shown in Figure 3d. There was no difference in survival between Grades 1 and 2; however, the T0 Grade 3 group had significantly poorer long-term survival compared with the combined Group 1 and Group 2 patients (median survival 6.5 years [95% CI 5.4 to 7.6] for no Grade 3, 3.8 years [2.2 to 4.9] for Grade 3, p ⫽ 0.0006, log rank test). P/F Grades 1 and 2 did not stratify patients into groups with differential long-term mortality at T0 or when grouped by worst score T(0 – 48). In addition, the predictive power of Grades 1 and 2 did not improve at T24 or T48 (data not shown). Therefore, we examined OI grade at T0 and as a worst score T(0 – 48) as a predictor of long-term survival. Patients classified as OI Grade 2 at T0 appeared to have an intermediate survival between Grade 1 and Grade 3; however, the differences in survival curves did not reach statistical significance and may have been due to chance (Figure 3e). Multivariate Analysis We analyzed T(0 – 48) Grade 3, T0 Grade 3 and T0 OI as predictors of outcome in multivariate models with other potential donor and recipient risk factors. The hazards of variables, such as need for cardiopulmonary bypass during the transplant operation or the development of Grade 3 graft dysfunction, are not necessarily constant over time (being greater in the peri-operative period); therefore, early outcome was modeled by logistic regression using 90-day death as the end-point, and late survival by a Cox proportional hazards analysis of peri-operative survivors. Only clinically meaningful variables associated with the respective end-points in a bivariate analysis at the p ⫽ 0.1 level were included in the models. Results of the analysis are shown in Table 4. Secondary Outcome Measures We evaluated the PGD grading system as a predictor of clinical outcomes apart from overall mortality, including patients’ length of ICU and hospital stay, as shown in Table 5. Patients who developed Grade 3 PGD by P/F score within 48 hours of surgery had significantly longer ICU and hospital stays vs patients classified as Grades 1 or 2. This was true regardless of whether SLT or BLT was performed. Of the 131 patients who continued to require mechanical ventilation at T48, 83 patients (63%) had Grade 3 PGD as their worst grade T(0 – 48), as compared with 35 patients (27%) with
376
Prekker et al.
The Journal of Heart and Lung Transplantation April 2006
Figure 3. (a) Long-term survival of the overall group. The numbers of patients at time of transplant and at the 5-year and 10-year time-points are listed at the bottom. The 5-year actuarial survival rate was 55%, and at 10 years it was 24%. (b) Overall survival of the 394 patients with PaO2/FIO2 data available stratified by worst P/F score T(0 – 48) reflecting the highest modified ISHLT grade assigned between T0 and T48. The numbers of patients at the time of transplant and at the 5-year and 10-year time-points are listed at the bottom. (c) Overall survival of the 348 patients who survived the 90-day peri-operative period, grouped by worst P/F score T(0 – 48). (d) Overall survival of the 394 patients grouped by P/F grade at T0. (e) Overall survival of the 389 patients with oxygenation index data available, grouped by T0 oxygenation index grade. OI, oxygenation index.
Grade 2 PGD, and only 13 patients (9%) were classified as Grade 1 PGD. Patients in the study who survived the peri-operative period were followed with pulmonary function testing at 1 year and 2 years after transplant. Means for forced expiratory volume in 1 second (FEV1) for SLT and BLT
patients are listed in Table 5, grouped by the worst P/F grade T(0 – 48). SLT patients had significantly poorer pulmonary function at both 1 and 2 years compared with patients who underwent BLT (Table 5). Increasing severity of PGD within the first 48 hours after lung transplantation was associated with decreasing pulmo-
The Journal of Heart and Lung Transplantation Volume 25, Number 4
Prekker et al.
Table 4. Multivariate Analysis No T(0–48) Grade 3 PGD BLT No CPB High pre-operative PA (mean) Recipient age Donor age Recipient male
p-value 0.03 ⬍0.01 ⬍0.01 0.04 NS NS NS
RR 0.85 0.61 0.67 1.02
95% CI 0.73–0.98 0.47–0.78 0.55–0.81 1.01–1.04
Significant univariate predictors of increased 90-day and long-term mortality are shown. Those variables that reached significance in our multivariate model are accompanied by p-values as well as relative risk and 95% confidence intervals. RR, relative risk; CI, confidence interval; SLT, single-lung transplant; CPB, cardiopulmonary bypass; PA, pulmonary arterial pressure.
nary function at both 1 and 2 years. However, differences among mean FEV1 values between those with and without severe PGD only reached statistical significance among BLT patients at 1 year (p ⫽ 0.002). DISCUSSION A useful system for grading PGD after lung transplantation must have several features. First, it should accurately stratify patients into groups with different perioperative death rates and perhaps different long-term survival rates. Second, it should rapidly identify patients who are developing the most severe forms of PGD, who might benefit from an effective intervention. Third, the measurement used to calculate the grade should be objective, reliably obtained, and simple to record, report and track within institutional databases and larger registries. The P/F grading system described in our work satisfies these requirements. The power of the PGD grading system (modified ISHLT criteria) to predict differential mortality is greatest when those patients with severe PGD (Grade 3, P/F ⬍200) are compared to those without severe PGD. Furthermore, Grade 3 PGD is an independent risk factor for poor long-term survival, when analyzed as the worst score T(0 – 48) in our multivariate model. Other inde-
377
pendent risk factors in our multivariate analysis were SLT, use of CPB, and increased pre-operative mean pulmonary artery pressure. The different components of the PGD definition are useful for different purposes. The T48 (and presumably T72) time-points are probably the end-points that best capture patients with the most significant forms of PGD. In our series, T48 demonstrated the clearest differentiation of 90-day death rates between mild PGD (Grade 1, 7%), moderate (Grade 2, 12%) and severe graft dysfunction (Grade 3, 33%) (p ⬍ 0.0001). The T0 P/F measurement is the best way to rapidly identify patients who might be destined to have a poor outcome and who might benefit from intervention. Additional early trend analysis might more accurately stratify this group of patients with respect to short-term mortality because one would expect that a patient with a rapidly improving ratio would have a more favorable prognosis than the overall group. Conversely, a patient with an unfavorable 12-hour change in their P/F ratio would be expected to have a higher operative mortality rate. The T0 data set is also useful in the sense that it is the most complete because nearly all transplant patients are mechanically ventilated when they arrive in the ICU. Grouping extubated patients in the favorable category at later time-points might bias the otherwise objective measurement of the P/F ratio. There was no obvious advantage to grouping patients by OI score, which incorporates airway pressures in addition to gas exchange into its definition.14 The qualitative stratification of short- and medium-term outcomes of Grade 1 and Grade 2 patients was improved as compared with the P/F grade, but the differences in curves did not reach the level of statistical significance. Our study has several limitations, the most obvious of which is that we did not strictly adhere to the ISHLT working group criteria for the assignment of grades. We did not collect data out to the 72-hour time-point nor did we analyze chest X-rays for the presence or absence
Table 5. Secondary Outcomes: ICU and Hospital Length of Stay, Pulmonary Function Tests ICU days Hospital days 1-year FEV1 SLT (n ⫽ 163) BLT (n ⫽ 103) 2-year FEV1 SLT (n ⫽ 130) BLT (n ⫽ 87)
Grade 1 4.4 ⫾ 1.0 14 ⫾ 1.5
Grade 2 5.4 ⫾ 1.1 16 ⫾ 1.7
Grade 3 12 ⫾ 1.0a 25 ⫾ 1.5a
p ⬍ 0.0001 p ⬍ 0.0001
58 ⫾ 1.7 92 ⫾ 4.3
52 ⫾ 1.9 87 ⫾ 4.4
52 ⫾ 1.8 71 ⫾ 4.5b
p ⫽ 0.056 p ⫽ 0.002
56 ⫾ 2.0 93 ⫾ 5.2
52 ⫾ 2.3 85 ⫾ 5.6
51 ⫾ 2.1 75 ⫾ 5.4
NS p ⫽ 0.059
Secondary outcomes are shown grouped by worst P/F grade T(0 – 48). ICU and hospital length of stay are listed as mean number of days. Patients were followed routinely with pulmonary function testing. Values at 1 and 2 years after transplant are reported as the percent of the predicted value based on the height and weight of the patient. FEV1, forced expiratory volume in 1 second; SLT, single-lung transplant; BLT, bilateral single-lung transplant. a p ⬍ 0.0001 vs Grades 1 and 2 PGD. b p ⫽ 0.002 vs Grades 1 and 2 PGD.
378
Prekker et al.
of infiltrates. We would expect that the presence of Grade 3 PGD at the 72-hour measurement would be predictive of an even greater peri-operative mortality rate than the 48-hour measurement. The T72 value might also correlate more with longer-term survival and with pulmonary function than the values that we analyzed. With regard to the chest X-rays, we found that most lung transplant patients do not have entirely clear films in the first 24 hours after transplant. With our institution’s change to electronic imaging, almost all of the ⬎2000 X-rays that would require analysis are either no longer available or are stored in warehouse archives. A clear chest X-ray 48 or 72 hours after a lung transplant might occur more frequently than it does in the earlier post-operative period and could enhance the predictive value of the ISHLT scoring system. However, this piece of recorded information is more subjective than the P/F ratio and would probably not be reliably reported to the registries by all centers. In conclusion, the development of severe (ISHLT Grade 3) primary graft dysfunction is associated with high operative mortality and decreased long-term survival. In the sub-set of patients who undergo bilateral single-lung transplantation, Grade 3 PGD is also associated with reduced pulmonary function at 1 and 2 years. In a multivariate analysis, risk factors for death in addition to Grade 3 PGD were the need for CPB during the transplant, having undergone single-lung transplant, and increasing pre-operative pulmonary artery pressure. Our results support the vision of the ISHLT working group to use the grading system as a standard tool in clinical care and in clinical trials to improve the outlook for patients undergoing pulmonary transplantation. Early identification of patients with the most severe forms of PGD may allow for the use of selective management strategies such as altered ventilation to reduce airway pressure as in ARDS, beta-blockade to limit perfusion of already injured lungs, pharmacologic agents to reduce ischemia–reperfusion injury,15–17 or even extracorporeal membrane oxygenation to allow the lungs a period of low-perfusion recovery for the most severe forms of PGD.3,18 REFERENCES 1. Trulock EP, Edwards LB, Taylor DO, et al. The registry of the International Society for Heart and Lung Transplantation: twenty-first official adult lung and heart–lung transplant report—2004. J Heart Lung Transplant 2004;23: 804 –15. 2. Christie JD, Kotloff RM, Ahya VN, et al. The effect of primary graft dysfunction on survival after lung transplantation. Am J Respir Crit Care Med 2005;171:1312– 6.
The Journal of Heart and Lung Transplantation April 2006
3. Dahlberg PS, Prekker ME, Herrington CS, Hertz MI, Park SJ. Medium-term results of extracorporeal membrane oxygenation for severe acute lung injury after lung transplantation. J Heart Lung Transplant 2004;23:979 – 84. 4. Christie JD, Bavaria JE, Palevsky HI, et al. Primary graft failure following lung transplantation. Chest 1998;114: 51– 60. 5. Christie JD, Kotloff RM, Pochettino A, et al. Clinical risk factors for primary graft failure following lung transplantation. Chest 2003;124:1232– 41. 6. Christie JD, Sager JS, Kimmel SE, et al. Impact of primary graft failure on outcomes following lung transplantation. Chest 2005;127:161–5. 7. Fiser SM, Kron IL, Long SM, Kaza AK, Kern JA, Tribble CG. Early intervention after severe oxygenation index elevation improves survival following lung transplantation. J Heart Lung Transplant 2001;20:631– 6. 8. Khan SU, Salloum J, O’Donovan PB, et al. Acute pulmonary edema after lung transplantation: the pulmonary reimplantation response. Chest 1999;116:187–94. 9. King RC, Binns OA, Rodriguez F, et al. Reperfusion injury significantly impacts clinical outcome after pulmonary transplantation. Ann Thorac Surg 2000;69:1681–5. 10. Thabut G, Vinatier I, Stern JB, et al. Primary graft failure following lung transplantation: predictive factors of mortality. Chest 2002;121:1876 – 82. 11. 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. 12. Christie JD, Carby M, Bag R, Corris P, Hertz M, Weill D. 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 –9. 13. Dahlberg PS, Prekker ME, Hertz MI, Thompson DJ, Park SJ. Recent trends in lung transplantation: the University of Minnesota experience. Clin Transplants 2002;XX:243–51. 14. Green TP, Timmons OD, Fackler JC, et al. The impact of extracorporeal membrane oxygenation on survival in pediatric patients with acute respiratory failure. Pediatric Critical Care Study Group. Crit Care Med 1996;24:323–9. 15. de Perrot M, Liu M, Waddell TK, Keshavjee S. Ischemia– reperfusion-induced lung injury. Am J Respir Crit Care Med 2003;167:490 –511. 16. Meade MO, Granton JT, Matte-Martyn A, et al. A randomized trial of inhaled nitric oxide to prevent ischemia– reperfusion injury after lung transplantation. Am J Respir Crit Care Med 2003;167:1483– 89. 17. Keshavjee S, Davis RD, Zamora MR, de Perrot M, Patterson GA. A randomized, placebo-controlled trial of complement inhibition in ischemia–reperfusion injury after lung transplantation in human beings. J Thorac Cardiovasc Surg 2005;129:423– 8. 18. Oto T, Rosenfeldt F, Rowland M, et al. Extracorporeal membrane oxygenation after lung transplantation: evolving technique improves outcomes. Ann Thorac Surg 2004; 78:1230 –5.