A Scheme for Defining Cause of Death and Its Application in the T Cell Depletion Trial

Article Outline

• Abstract
• Introduction
• Methods
  • Patients
  • Infection Prophylaxis
  • Review Panel and COD Reporting Rules
  • Statistical Analysis
• Results
  • Patients
  • Primary COD
  • Comparison With Causes of Death Reported to the CIBMTR
• Discussion
• Acknowledgments
• References
• Copyright

Abstract 

The primary cause of death (COD) provides important information in many studies of hematopoietic stem cell transplantation (HSCT). A panel of experts critically assessed the CODs submitted by 15 transplantation centers for 281 patients who died in a randomized multicenter trial of unrelated HSCT. The panel reviewed the CODs reported by the transplantation centers, which used the Center for International Blood and Marrow Transplant Research and National Marrow Donor Program COD reporting form. The panel determined that the existing criteria for primary and contributing CODs lacked sufficient stringency for uniform interpretation. A hierarchy was developed and applied to the T cell depletion project. Using its scheme, the panel reclassified 157 CODs (56%) reported by the transplantation centers. The changes resulted in increased recognition of graft-versus-host disease as the primary COD and a concomitant decrease in attribution of the primary COD to infection. This algorithm promotes consistent assignment of primary and contributing CODs for patients with leukemia or lymphoma who expire after myeloablative allogeneic HSCT.

Key Words: Cause of death, Hematopoietic stem cell transplantation, T cell depletion

Back to Article Outline

Introduction 

The cause of death (COD) provides important information on many patient cohorts, including those participating in clinical studies of hematopoietic stem cell transplantation (HSCT). Analysis of COD data can help identify problems, modify clinical behavior, and guide new studies.

Clinical studies of HSCT commonly report COD. The standard form (Form 190) from the Center for International Blood and Marrow Transplant Research and National Marrow Donor Program (CIBMTR/NMDP) lists common causes but provides limited guidance to distinguish among competing CODs, such as graft-versus-host disease (GVHD) and infection. Differences in reporting practices compromise clinical studies of HSCT and prevent reliable comparisons of results between studies.

COD was an important component of a large prospective multicenter trial of unrelated donor HSCT in which recipients were randomized to receive unmodified marrow and posttransplantation immunosuppression with methotrexate (MTX) and cyclosporine (CSA) or T cell–depleted (TCD) marrow and posttransplantation immunosuppression with CSA alone for prevention of GVHD [1]. The primary objective of this TCD trial was to compare the 3-year disease-free survival (DFS) of the 2 groups; secondary objectives included comparisons of (1) overall survival (OS), (2) relapse, (3) engraftment, (4) severity and incidence of GVHD, (5) severity and incidence of infections, and (6) regimen-related toxicity and adverse events. The CODs were reported by transplantation centers using CIBMTR/NMDP COD codes. The reported CODs were then reviewed by an expert panel with access to relevant medical records, including infectious complications that had been reviewed and audited by a panel of experts in this area. The COD panel recognized substantial variation between centers in assigning primary CODs. The panel used the same primary COD list from CIBMTR/NMDP Form 190 but defined a hierarchy for determination of COD, and then applied this scheme to the study patients.

Back to Article Outline

Methods 

Patients 

A total of 410 patients with lymphohematopoietic malignancies were randomized to receive TCD marrow and CSA after transplantation (TCD arm; n = 203) or unmodified marrow with MTX and CSA (M/C arm; n = 207) for GVHD prophylaxis. The method used for T-cell depletion was elutriation in 67 patients and monoclonal antibody (mAb) and complement lysis in 136 patients. Accrual occurred between March 1995 and October 2000 at 15 centers in the United States. Minimum follow-up was 18 months. Institutional review board approval was obtained from each center. Written consent was obtained from each patient or legal guardian before randomization. The cutoff date for analysis was April 2002.

Infection Prophylaxis 

For transplantation, patients were hospitalized in single reverse-isolation rooms. Each center had guidelines for preventing fungal infections, herpes simplex and zoster virus reactivation, Pneumocystis pneumonia and cytomegalovirus (CMV) reactivation after engraftment, and infection with encapsulated Gram-positive organisms during treatment of GVHD.

Review Panel and COD Reporting Rules 

Separate panels reviewed weekly GVHD staging assessments and infections. The acute GVHD (aGVHD) panel developed a computer algorithm using the consensus grading scheme [2]. GVHD occurring before day 100 was classified as acute, and that occurring after day 100 was classified as chronic. The infection review panel developed an algorithm to categorize infection and tallied the numbers of severe, life-threatening, and fatal infections per patient. A severity level of fatal was assigned if the infection was present at the time of death or if it contributed significantly to death.

After the aGVHD and infection review panels completed their evaluations, an expert panel composed of 6 transplantation physicians, an infectious disease specialist, and a statistician reviewed the COD reports for every patient during weekly conference calls over a 10-month period after the close of the study. The COD form used by the CIBMTR/NMDP (Form 190) served as the template for reporting COD by the transplantation center. Autopsy records were reviewed. The panel was blinded to treatment arm.

For every patient, the panel reviewed a demographic summary that included age, disease, and disease stage; days to neutrophil and platelet engraftment and chimerism results; maximum organ toxicity within 28 days of transplantation using the Bearman scale [3]; posttransplantation complications, such as bronchiolitis obliterans organizing pneumonia (BOOP), adult respiratory distress syndrome (ARDS), and veno-occlusive disease; weekly aGVHD scoring records; presence and stage of chronic GVHD (cGVHD); immunosuppressive medications at the time of death; documentation of relapse; all infections, including dates and severity; and the transplantation center–designated COD. The COD was assigned by reaching consensus on a conference call through case presentation and discussion.

A flow diagram illustrating the hierarchy developed by the panel for assigning primary and contributing CODs is depicted in Figure 1. If relapse occurred, then this was assigned as the primary COD. Primary graft failure was defined as failure to achieve an absolute neutrophil count (ANC) of 500/mm³ by day 28 after transplantation. For patients who died before day 14 after transplantation, graft failure was not listed as the primary COD. For patients with aGVHD who required systemic immunosuppressive therapy before day 28 after transplantation and also experienced primary graft failure, aGVHD was identified as the primary COD, with graft failure as the secondary COD. Patients with a primary COD of either graft failure or GVHD had contributing causes (eg, infection, organ failure) recorded as secondary or tertiary. If relapse, graft failure, or GVHD were present at the time of death, then this was assigned as the primary COD. In this setting, infections and organ failure were recorded as secondary and tertiary CODs. If relapse was detected at autopsy for the first time, then relapse was recorded as a secondary COD rather than as the primary COD.

• 
  • View Large Image
  • Download to PowerPoint
• 
  • View Large Image
  • Download to PowerPoint
• 
  • View Large Image
  • Download to PowerPoint

• Figure 1 

  Hierarchy for COD.

If none of the aforementioned causes were recorded as the primary COD, then the algorithm continued with infection and organ failure, according to Figure 1.

Statistical Analysis 

All contingency tables and χ² tests were performed using SAS version 8.2 (SAS Institute, Cary, NC).

Back to Article Outline

Results 

Patients 

Of the 410 patients, 5 died before undergoing transplantation. Seven patients assigned to receive TCD received M/C instead, because of low cell dose (n = 5), logistics of total body irradiation (n = 1), or complications of the conditioning regimen (TBI; n = 1). Except for infused cell dose, baseline characteristics were similar for the 2 treatment arms (data not shown) [1].

Primary COD 

Death occurred in 141 of the 203 TCD patients (69%) and in 140 of the 207 M/C patients (68%). A total of 70 patients were autopsied. In 56% of the patients, the primary COD assigned by the reporting transplantation center was changed by the review panel, using the diagram shown in Figure 1. The difference between treatment arms in the number of patients whose primary COD was changed was not statistically significant (57% for the TCD group and 54% for the unmodified group).

For the 157 patients in whom a change was made, the differences between the primary COD assigned by the panel and that reported by the transplantation center are shown in Table 1. Determination that aGVHD (n = 48) or cGVHD (n = 41) was the primary COD accounted for the majority of the changes.

Table 1. Differences in panel review COD and transplantation center–reported COD

GVHD (acute, 24%; chronic, 19%) was the most common COD designated by the panel. There was no difference in the incidence of aGVHD as the primary COD (TCD, 22%; M/C, 26%) between treatment arms, although the incidence of grade III/IV aGVHD differed significantly between treatment arms (TCD, 18% vs M/C, 37%; P < .001) [1]. The incidence of cGVHD as the primary COD was not different by treatment arm (TCD, 18% vs M/C, 19%) [4]. The secondary causes of death for patients in whom GVHD was the primary COD are given in Table 2. There were no significant differences between treatment arms. Viral infections accounted for 4% of the primary CODs in both treatment arms; however, viral infections after transplantation were more common in the TCD arm [2]. There was no difference between the treatment arms for any of the primary CODs (Table 3).

Table 2. Secondary COD by treatment in patients in whom the primary COD was GVHD

	TCD (n = 141)		Unmodified (n = 140)		Total (n = 281)
Primary COD	n	(%)	n	(%)	n	(%)
Acute GVHD
Secondary CODs	31	(22)	36	(26)	67	(24)
Primary graft failure	0		1		1
Infection	26		18		44
Organ failure	3		9		12
Hemorrhage	0		1		1
None	2		7		9
Chronic GVHD
Secondary CODs	26	(18)	27	(19)	53	(19)
Infection	12		13		25
Organ failure	2		2		4
Secondary malignancy	1		0		1
Other	2		1		3
None	9		11		20

Table 3. Primary COD by treatment

	TCD (n = 141)		Unmodified (n = 140)		Total (n = 281)
Primary COD	n	(%)	n	(%)	n	(%)
Recurrence/relapse	31	(22)	30	(21)	61	(22)
Persistent disease	2	(1)	0	(0)	2	(1)
Graft failure
Primary	8	(6)	12	(9)	20	(7)
Autologous recovery	1	(1)	0	(0)	1	(0)
Rejection	1	(1)	0	(0)	1	(0)
GVHD
Acute	31	(22)	36	(26)	67	(24)
Chronic	26	(18)	27	(19)	53	(19)
Infection
Fungal	5	(4)	2	(1)	7	(2)
Bacterial	1	(1)	3	(2)	4	(1)
Viral	6	(4)	5	(4)	11	(4)
Other1	6	(4)	3	(2)	9	(3)
Organ failure
Other2	4	(3)	8	(6)	12	(4)
Pulmonary	8	(6)	4	(3)	12	(4)
Non-ARDS	3		2		5
ARDS	5		2		7
Cardiac	4	(3)	1	(1)	5	(2)
Liver	1	(1)	2	(1)	3	(1)
Secondary graft failure	2	(1)	2	(1)	4	(1)
Secondary malignancy	0	(0)	1	(1)	1	(0)
Hemorrhage	3	(2)	1	(1)	4	(1)
Other3	1	(1)	3	(2)	4	(1)

Notable modifications to the assigned primary COD after panel review occurred with assignment of 6 cases of relapse, 2 cases of persistent disease, 15 cases of primary graft failure, 51 cases of aGVHD, and 41 cases of cGVHD. There were no differences by treatment arm.

Comparison With Causes of Death Reported to the CIBMTR 

Figure 2 compares the primary CODs reported to the CIBMTR for 2896 patients undergoing unrelated donor transplantation during a similar time period, including patients who received TCD grafts. The form for reporting COD was the same as that used by the transplantation centers in this trial. The pie charts for the CIBMTR data (A) are nearly identical to the CODs reported by transplantation centers in this trial (B); the revised assignment based on the hierarchy developed by the COD committee (C) differs markedly. Notable findings include increases in GVHD as the primary COD (with a concomitant decrease in infection) and the almost complete absence (1 case) of idiopathic pneumonia (IP) as the primary COD.

• 
  • View Large Image
  • Download to PowerPoint

• Figure 2 

  COD after unrelated donor transplantation.

Back to Article Outline

Discussion 

Despite its recognized importance, stringent rules for assignment of COD after myeloblative HSCT have not been developed. Guidance is required because multiple factors often contribute to death after allogeneic transplantation 5, 6. The immediate COD may directly result from a condition (eg, graft failure, GVHD) that is responsible for the fatal outcome. Inconsistencies in the designation of COD by individual investigators and between investigators and centers is problematic. Figure 2 illustrates that the COD designated by the centers in this trial reflects that for all patients reported to the CIBMTR.

This report describes the development and use of a scheme for determining COD in patients with lymphohematopoietic malignancies who underwent myeloablative conditioning and unrelated transplantation. Data were collected prospectively from multiple transplantation centers and audited for primary and secondary CODs, weekly GVHD assessments, and documentation of all infectious episodes. In a majority of patients, the COD reported by the transplantation centers was reassigned by the review panel. The newly designated primary COD and the number of changes were similar between the 2 treatment arms.

Recurrence or persistence of underlying hematologic disease is given precedence. In the proposed schema, a patient who relapses, receives additional treatment, and dies of a fungal infection is assigned recurrence of disease as the primary COD. Fungal infection is considered a secondary cause.

If there is no evidence of relapse, graft failure is the next most important cause, followed by GVHD. The pie charts in Figure 2 illustrate that the application of this series of rules resulted in a significant shift between the primary COD reported by the transplantation centers and the final decision reached by the review panel, particularly as related to GVHD. In a patient who died while receiving treatment for GVHD, GVHD was assigned as the primary COD, even if there was no clinical evidence of GVHD. GVHD was also the primary COD for patients who were treated for GVHD before day 28 and experienced primary graft failure.

Interstitial pneumonitis (IP) with a known underlying infectious cause was reclassified as an infection, whereas historically the CIBMTR would not list CMV IP as infection. This resulted in a decreased number of primary CODs attributed to IP. We recommend that IP should represent “idiopathic” rather than “interstitial” pneumonitis, because “interstitial” may include viral infections.

This hierarchy was developed for myeloablative HSCT, particularly for patients with leukemia or lymphoma. It has been used to analyze data obtained from the Cord Blood Transplantation (COBLT) study 7, 8, 9. Modified hierarchies for reporting primary CODs might be considered in other settings.

Correct delineation of the primary COD is important. The growing number of multi-institutional trials increases the need for uniform standards. The rules proposed herein encourage uniformity in the assignment of COD. This refined hierarchy places more importance on conditions preceding and responsible for the immediate COD and allows for prioritization of multiple problems. A similar approach has proved useful in other conditions; for example, an algorithm favoring linkage of human immunodeficiency virus/acquired immunodeficiency syndrome to COD proved important in recognizing the effectiveness of antiretroviral therapy [10]. The specificity of the scheme promotes consistency between different physicians and centers, permitting greater accuracy and reliable comparisons of data. Inaccurate reporting of the primary COD can lead to false assumptions and compromise future trials. It is important for the transplantation community to accept and incorporate a defined set of rules defining COD for future trials.

Back to Article Outline

Acknowledgments 

We thank the patients who willingly entered this large clinical trial; the physicians, nurses, and other support staff who cared for them during the transplant procedure; and the clinical investigation team at each participating institution for data collection and follow-up. We also thank the National Heart, Lung, and Blood Institute (NHLBI) for support of this trial. In addition to the authors, the following transplantation centers, study physicians, and experts contributed to this study: University of Minnesota (n = 103; Daniel J. Weisdorf, MD), Memorial Sloan-Kettering Cancer Center (n = 70; Richard O'Reilly, MD and Esperanza Papadoupolos, MD), Medical College of Virginia (n = 53), Wake Forest University-Baptist (n = 36), University of Nebraska (n = 34; Stephen Pavletic, MD and Michael Bishop, MD), University of Utah (n = 33; Patrick Beatty, MD and Finn Bo Petersen, MD), Stanford University (n = 25; Robert Negrin, MD), University of Iowa (n = 19; Robert Gingrich, MD), University of South Carolina (n = 13; Jean Henslee-Downey, MD and Kamar Godder, MD), Ohio State University (n = 6), Duke University (n = 6; Joanne Kurtzberg, MD), University of Kentucky (n = 5; John S. Thompson, MD and Gordon Phillips, MD), Medical College of Wisconsin (n = 4; Neal Flomenberg, MD), Western Pennsylvania Hospital (n = 2; Richard Shattuck, MD), University of Pittsburgh (n = 1; Albert Donnenberg, PhD), the EMMES Corporation (Donald Stablein, PhD and Elizabeth Wagner, M.P.H.), and the National Heart, Lung, and Blood Institute (LeeAnn Jensen, PhD, Nancy Geller, PhD, and Paul McCurdy, MD).

Back to Article Outline

References 

1. Wagner JE, Thompson JS, Carter SL, et al. Impact of graft-versus-host disease prophylaxis on 3 year disease free survival in recipients of unrelated donor bone marrow transplantation: results of a multi-centre, randomized Phase II–III trial. Lancet. 2005;366:9487
   • View In Article
2. Weisdorf DJ, Hurd D, Carter S, et al. Prospective grading of graft-versus-host disease after unrelated donor marrow transplantation: a grading algorithm versus blinded expert panel review. Biol Blood Marrow Transplant. 2003;9:512–518
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (178 KB)
   • CrossRef
3. Bearman SI, Appelbaum FR, Buckner CD, et al. Regimen-related toxicity in patients undergoing bone marrow transplantation. J Clin Oncol. 1988;6:1562–1568
   • View In Article
   • MEDLINE
4. Pavletic SZ, Carter SL, Kernan NA, et al. Influence of T-cell depletion on chronic graft-versus-host disease: results of a multi-center randomized trial in unrelated marrow donor transplantation. Blood. 2005;106:3308–3313
   • View In Article
   • MEDLINE
   • CrossRef
5. Hofmeister CC, Marinier DE, Czerlanis C, et al. Clinical utility of autopsy after hematopoietic stem cell transplantation. Biol Blood Marrow Transplant. 2007;13:26–30
   • View In Article
   • Full Text
   • Full-Text PDF (64 KB)
   • CrossRef
6. Van Burik JA, Leisenring W, Myerson D, et al. The effect of prophylactic fluconazole on the clinical spectrum of fungal diseases in bone marrow transplant recipients with special attention to hepatic candidiasis: an autopsy study of 355 patients. Medicine. 1998;77:246–254
   • View In Article
   • MEDLINE
   • CrossRef
7. Cornetta K, Laughlin M, Carter S, et al. Umbilical cord blood transplantation in adults: results of the prospective Cord Blood Transplantation (COBLT) Study. Biol Blood Marrow Transplant. 2005;11:149–160
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (242 KB)
   • CrossRef
8. Wall DA, Carter SL, Kernan NA, et al. Busulfan/melphalan/ATG followed by unrelated donor cord blood transplantation for treatment of infant leukemia and leukemia in young children: the COBLT experience. Biol Blood Marrow Transplant. 2005;11:637–646
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (381 KB)
   • CrossRef
9. Martin PL, Carter SL, Kernan NA, et al. Results of the Cord Blood Transplantation Study (COBLT): clinical outcomes of unrelated donor umbilical cord blood transplantation in pediatric patients with lysosomal and peroxisomal storage diseases. Biol Blood Marrow Transplant. 2005;12:184–194
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (415 KB)
   • CrossRef
10. Cohen MH, French AL, Benning L, et al. Causes of death among women with human immunodeficiency virus infection in the era of combination antiretroviral therapy. Am J Med. 2002;113:91–98
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (111 KB)
    • CrossRef