Survival Outcomes of Allogeneic Hematopoietic Cell Transplants with EBV positive or EBV negative Post Transplant Lymphoproliferative Disorder (PTLD), A CIBMTR Study.

Abstract

Background:

Post-transplant lymphoproliferative disorders (PTLD) are associated with significant morbidity and mortality following allogeneic hematopoietic cell transplant (alloHCT). Although most PTLD is EBV-positive (EBVpos), EBV-negative (EBVneg) PTLD is reported; yet its incidence and clinical impact remain largely undefined. Furthermore, factors at the time of transplant impacting survival following PTLD are not well described.

Methods:

Between 2002 and 2014, 432 cases of PTLD following alloHCT were reported to the Center for International Blood and Marrow Transplant Research (CIBMTR). After exclusions, 267 cases (EBVpos = 222, 83%; EBVneg = 45, 17%) were analyzed.

Results:

Two-hundred and eight patients (78%) received in vivo T-cell depletion (TCD) with either anti-thymocyte globulin (ATG) or alemtuzumab. Incidence of PTLD was highest using umbilical cord donors (UCB, 1.60%) and lowest using matched related donors (MRD, 0.40%). Clinical features and histology did not significantly differ among EBVpos or EBVneg PTLD cases except that absolute lymphocyte count recovery was slower and CMV reactivation was later in EBVneg PTLD [EBVpos 32 (5 – 95) days versus EBVneg 47 (10 – 70) days, p=0.016]. There was no impact on survival by EBV-status in multivariable analysis [EBVneg RR 1.42, 95% CI 0.94–2.15, p = 0.097].

Conclusions:

There is no difference in survival outcomes for patients with EBVpos or EBVneg PTLD occurring following alloHCT and 1-year survival is poor. Features of conditioning and use of serotherapy remain important.

INTRODUCTION:

Post-transplant lymphoproliferative disorder (PTLD) usually occurs within the first 6 months following allogeneic hematopoietic cell transplant (alloHCT) prior to effective reconstitution of cytotoxic T lymphocytes (CTL) needed to prevent Epstein Barr Virus (EBV)-mediated, B-cell transformation into lymphoblasts.^1–6 PTLD can be EBV negative (EBV^neg), potentially induced by other viral reactivations.⁵ Following alloHCT, EBV DNAemia can be detected in approximately 31% of unmanipulated T-cell replete and 65% T-cell depleted (TCD) graft recipients respectively.⁶ Incidence of PTLD is approximately 224, 54, and 31 per 100,000 transplants during the first, second, and sixth year following transplantation, respectively.^4,7 PTLD represents a heterogeneous group of non-Hodgkin’s lymphomas histologically classified as polymorphic or monomorphic with most early lesions being polymorphic. Monomorphic PTLD is further characterized as diffuse large B-cell lymphoma (DLBCL), Burkitt lymphoma, Burkitt-like lymphoma, and Hodgkin-like lymphoma.⁴

Known risk factors for EBV^pos PTLD include: a high degree of HLA mismatch; ex vivo or in vivo T-cell depletion (using anti-thymocyte globulin (ATG) or alemtuzumab); second alloHCT; older age of recipient; and the intensity and duration of immunosuppression used for graft-versus-host-disease (GvHD) prophylaxis or treatment.^3,8–13 However, the current literature lacks information on outcomes of as well as the influence of ATG or alemtuzumab on EBV^pos or EBV^neg PTLD.^1,5,7–9,13 Our understanding is extrapolated from solid organ transplant associated PTLD⁷, therefore, we interrogated the observational database of the Center for International Blood and Marrow Transplantation (CIBMTR) to analyze clinical outcomes for patients who developed EBV^pos and EBV^neg PTLD following alloHCT in the contemporary transplant era.

MATERIAL AND METHODS:

Data source:

The CIBMTR is a working group of more than 400 transplantation centers worldwide that contribute detailed data on HCT to a statistical center at the Medical College of Wisconsin. Participating centers are required to report all transplantations consecutively with longitudinal follow-up. On-site audits monitor data and reporting compliance. Computerized checks for discrepancies, physicians’ review of submitted data and on-site audits of participating centers ensure data quality. Observational studies conducted by the CIBMTR are performed in compliance with all applicable federal regulations pertaining to the protection of human research participants. The CIBMTR collects data at two levels: Transplant Essential Data (TED) and Comprehensive Report Form (CRF) data. TED data include disease type, age, gender, pre-alloHCT disease stage and chemotherapy-responsiveness, date of diagnosis, graft type, conditioning regimen, post-transplant disease progression and survival, development of a new malignancy and cause of death. All CIBMTR centers contribute TED data. More detailed disease and pre- and post-transplant clinical information, including infection related data, are collected on a subset of registered patients selected for CRF data by a weighted randomization scheme. TED and CRF level data are collected pretransplant, 100 days and six months post-HCT and annually thereafter or until death. Only CIBMTR CRF data are used in this analysis and all patients have signed consents for CIBMTR registry data collection.

Patients and definitions

Centers report development of secondary malignancies, including PTLD as a yes/no variable, a subsequent yes/no variable for EBV positivity, and the date of onset. The CIBMTR does not capture data regarding use of immune suppression specifically at the time of PTLD diagnosis nor treatment or outcomes data specific to the PTLD. Amongst alloHCT reported to the CIBMTR between 2002 and 2014 (n=41,410), 432 alloHCT patients were reported with PTLD. Of the 432 identified, patients were excluded from final analyses due to initial diagnosis of lymphoma or other B cell malignancies (n=51), diagnosis of immunodeficiency or autoimmune disorders (n=16), twin or multiple donors (n=8), lack of follow-up and other incomplete data (n=90). Therefore, the analyzed population includes 267 cases of PTLD from 106 centers. After applications of similar exclusions to the non-PTLD alloHCT patients, 26,443 patients remained.

Pathology reports were requested for all cases and centers provided reports for central review on 167 cases (63%) who developed PTLD. EBV^pos PTLD was defined based on the reported histologic finding of EBV-encoded small RNAs (EBERs) documented in the pathology report. Similarly, negative immunostains and/or negative EBV PCR defined EBV^neg PTLD. A comparison of baseline characteristics and outcomes for patients with and without pathology reports [Pathology report available EBV^pos=135, 61%; EBV^neg=32, 71%] were similar (data not shown). Therefore, all 267 reported PTLD cases after alloHCT were included in the final analysis.

Conditioning intensity and HLA categories utilized previously published definitions^14–16. Disease status was categorized as early [AML/ALL in complete remission (CR) 1, = 65; MDS with <5% blasts or isolated 5q- syndrome, = 13; CML in CR1 or chronic phase (CP) 1, = 4], intermediate [ALL in CR2+, = 32; CML in CR2/CP2, = 4; plasma cell disorder, =1] and advanced [AML/ALL with active disease =25; MDS with ≥5% blasts/MPS, = 37]. Few patients (AML/ALL =4, CML =2) had an unknown disease status and the remaining patients had non-malignant diseases.

Study Endpoints

The study aimed to describe the overall survival (OS) of patients from the diagnosis of EBV^pos and EBV^neg PTLD. Additional, secondary endpoints included time of onset of PTLD after transplant and identification of any risk factors present at the time of transplant that may affect survival after diagnosis with PTLD. Risk factors for development of PTLD were not examined.

Statistical analysis

Patient demographics and transplant-related information were analyzed using the Chi-square test, Fisher’s exact test or the Kruskal Wallis test, as appropriate. The Kaplan-Meier product-limit estimator estimated the median and range of the follow-up time. Survival probability was calculated using the Kaplan-Meier estimator, with the variance estimated by Greenwood’s formula. Cox regression modeling was applied to analyze OS from the time of PTLD onset. The proportional hazards assumption was tested and violations added as time-dependent covariates. Interactions between significant covariates were investigated. The main effect variable for the OS model was the EBV-status of the PTLD. Due to interactions between disease (malignant vs. non-malignant), use of TBI, and conditioning intensity, a composite variable for these factors was used in the multivariable Cox model [malignant disease with myeloablative conditioning (MAC) with total body irradiation (TBI) vs. malignant disease with MAC and no TBI vs malignant disease with non-myleoablative/reduced intensity conditioning (NMA/RIC) with TBI vs. malignant disease with NMA/RIC and no TBI vs. non-malignant disease with TBI vs. non-malignant disease and no TBI]. Additional transplant variables examined included age [0 – 20 years vs. 21 – 50 years vs. ≥51 years], donor type [related vs. unrelated vs. cord], and in vivo T-cell depletion [none vs. ATG/alemtuzumab]. Time-dependent variables occurring after transplant but prior to PTLD diagnosis were examined including CMV viremia [no vs. yes], EBV viremia [no vs. yes] and GvHD [no vs. yes].

RESULTS:

Patient and PTLD characteristics

After applying exclusion criteria, 267 PTLD cases were identified in 26,710 alloHCT patients reported during the study period, resulting in a 1% reported incidence of PTLD. PTLD occurrence varied by relationship between donor and recipient and HLA match as follows: 0.40% with matched related donor (MRD); 1.41% with mismatched related donors; 1.16% matched unrelated donor (MUD); 1.47% mismatched unrelated donor (MMUD), and 1.60% with UCB. An additional 13 unrelated donors did not have sufficient HLA typing to categorize as MUD or MMUD.

Table 1 describes the patient characteristics for the EBV^pos PTLD (n=222) and EBV^neg PTLD (n = 45) populations. There was no difference between EBV^pos and EBV^neg groups. Median age was 32 (<1–76) years and 47 (1–70) years (p=0.2) for EBV^pos and EBV^neg patients, respectively. The majority of patients with PTLD were Caucasian (79%), male (60%), and had AML/MDS (58%). For patients with malignant disease, myeloablative (MA) conditioning (69%) predominated. Approximately 46% of patients received total body irradiation. The majority of patients received allografts from donors that were CMV seronegative (n=187, 70%). Donor EBV serostatus was available for 120 (45%) donors and only 34 (28%) were sero-positive. Fifty-three EBV seropositive recipients received grafts from an EBV seronegative donor (EBV^pos PTLD = 43 and EBV^neg PTLD = 10). The use of T-cell depletion (TCD), either ex vivo or in vivo with ATG or alemtuzumab, was similar in the cohorts and 71% of patients received ATG (n = 190).

Table 1:

Characteristics of patients with PTLD, reported to the CIBMTR, from 2002 to 2014

Variable	EBVpos PTLD N = 222 N(%)	EBVneg PTLD N = 45 N(%)	p-value
Number of centers	94	35
Gender, male	133 (60)	27 (60)	0.991
Karnofsky performance @ alloHCT ≥90	167 (75)	29 (64)	0.105
Time from diagnosis to TX, median (range), months	9 (<1 – 327)	8 (1 – 150)	0.287
Age at transplant, years			0.424
0–20	84 (38)	14 (31)
21–40	51 (23)	7 (16)
41–60	57 (26)	15 (33)
>60	30 (14)	9 (20)
Disease			0.751
AML	87 (39)	18 (40)
ALL	18 ( 8)	3 ( 7)
CML	8 ( 4)	2 ( 4)
MDS	40 (18)	10 (22)
Other Malignant disease	3 ( 1)	0
Severe aplastic anemia	46 (21)	6 (13)
Inherited non-malignant disease	20 ( 9)	6 (13)
HCT-CI			0.076
0	55 (25)	5 (11)
1 – 2	33 (15)	7 (16)
≥3	36 (16)	13 (29)
Missing	98 (44)	20 (44)
Conditioning regimen intensity			0.122
Myeloablative	103 (46)	25 (56)
RIC/NMA	51 (23)	7 (15)
Non-Malignant diseases	68 (31)	12 (27)
Missing	0	1 ( 2)
Donor-recipient HLA match			0.915
HLA identical sib	32 (14)	9 (20)
Mismatched other related	9 ( 4)	1 ( 2)
Unrelated, well matched	72 (32)	14 (31)
Unrelated partially matched	30 (14)	4 ( 9)
Unrelated mismatched	7 ( 3)	1 ( 2)
Unrelated HLA missing	11 ( 5)	2 ( 4)
Cords	61 (27)	14 (31)
Donor-recipient sex match			0.951
Cords	61 (27)	14 (31)
Female-Male	28 (13)	5 (11)
Other gender match	131 (59)	26 (58)
Missing	2 (<1)	0
Donor-recipient CMV status			0.632
Negative/Negative	73 (33)	16 (36)
Any Positive	146 (66)	28 (62)
Missing	3 ( 1)	1 ( 2)
Stem cell source			0.781
Marrow	65 (29)	11 (24)
Peripheral Blood	96 (43)	20 (44)
Cord blood	61 (27)	14 (31)
ATG/alemtuzumab as conditioning/GVHD prophylaxis			0.569
ATG + CAMPATH	1 (<1)	0
ATG alone	155 (70)	34 (76)
CAMPATH alone	17 ( 8)	1 ( 2)
No ATG or CAMPATH	49 (22)	10 (22)
GVHD prophylaxis*			0.384
Ex vivo T-cell depletion	17 ( 8)	2 ( 4)
CD34 selection	10 ( 5)	2 ( 4)
Post-transplant Cyclophosphamide	2 (<1)	1 ( 2)
TAC/CSA + MMF +- others	62 (28)	16 (36)
TAC/CSA + MTX +- others	84 (38)	14 (31)
TAC/CSA + others	27 (12)	9 (20)
TAC/CSA alone	16 ( 8)	1 ( 2)
Other GVHD prophylaxis	4 ( 2)	0
Year of transplant			0.686
2002 – 2004	31 (15)	9 (20)
2005 – 2007	65 (29)	11 (24)
2008 – 2010	66 (29)	13 (28)
2011 – 2014	60 (27)	12 (27)

Abbreviations: ALL=acute lymphoblastic leukemia; AML=acute myelogenous leukemia; MDS =myelodysplastic syndromes; HLA=human leukocyte antigen; TAC= tacrolimus; CSA = cyclosporine; MTX = methotrexate; TBI = total body irradiation; MMF=mycophenolate mofetil; CSA= cyclosporine A; ATG=anti-thymocyte globulin; NMA: Nonmyeloablative; RIC : Reduced intensity conditioning²².

^*14 patients received mTOR inhibitors

PTLD developed at a median of 3 months (range <1–102) post-alloHCT in patients with EBV^pos PTLD and 5 months (range 1–110) in patients with EBV^neg PTLD (p=0.1). Median reported white blood cell (WBC) count at 3 months post-alloHCT was 4×10⁹/L (range <1–39) and 5×10⁹/L (range <1–55) for EBV^pos and EBV^neg PTLD, respectively (p=0.016). The median absolute lymphocyte count (ALC) at this time was 1.16×10⁹/L (range, <1–21) and 0.65×10⁹/L (range, <1–20) for EBV^pos and EBV^neg PTLD, respectively (p=0.001). Use of intravenous immunoglobulin (IVIG) at any time post-transplant was reported in 108 patients (EBV^pos = 98; EBV^neg = 16); 35 reported the indication for IVIG as infection (CMV = 15, RSV = 1, unspecified infection = 20). CMV DNAemia prior to PTLD occurred in 58 (22%) patients. The median time to CMV DNAemia was 32 days (5–95) and 47 days (10–70) for EBV^pos and EBV^neg PTLD, respectively (p=0.016). A preceding EBV DNAemia occurred at a median of 60 days (range 17–100) in 75 (28%) patients. In the patients with EBV^neg PTLD, 10 (22%) reported a preceding EBV DNAemia but had pathologically confirmed EBV^neg PTLD on central review. Acute GVHD grade II-IV occurred in 83 patients prior to the diagnosis of PTLD [EBV^pos=72 (32%), EBV^neg =11 (24%)]. Overall, 43 patients with PTLD developed chronic GVHD [34 EBV^pos (15%), 9 (20%) EBV^neg PTLD] prior to developing PTLD.

Immunopathology and clonality

Pathology reports (Table 2) were available for 167 PTLD cases [135 (61%) EBV^pos; 32 (71%) EBV^neg)] and showed infrequent T-cell (n=3; 2 EBV^pos, 1 EBV ^neg) and Burkitt lymphoma subtypes (n=3; 1 EBV^pos, 2 EBV^neg). Histopathology subtypes were comparable between the two PTLD groups for DLBCL, polymorphic PTLD, PTLD NOS and plasmacytoid morphology. Specific immunohistochemical staining varied with frequent missing data. The majority of EBV^pos PTLD cases were monomorphic (65%, n=88), and Ki-67 was >80% positive in 22% (n = 7/30). Immunostains were positive for CD20 in 74% (n=100), Bcl-2 in 20% (n=27), Bcl-6 in 10% (n=14), and CD-30 in 25% (n=34). Similarly, EBV^neg PTLD was monomorphic in 66% (n=21) patients, and Ki-67 staining was >80% positive in 25% (n=8). For EBV^neg PTLD, immunostains were positive for CD20 in 59% (n=19), Bcl-2 in 28% (n=9), Bcl-6 in 13% (n=4) and CD30 in 28% (n=9).

Table 2:

Immunopathology of 167 PTLD cases in which pathology reports available

Variable	EBV positive [N(%)]	EBV negative [N(%)]	p-value
Number of patients	135	32
Morphologic Type			0.188
Monomorphic	88 (65)	21 (66)
Polymorphic	38 (28)	6 (19)
Missing	9 ( 7)	5 (16)
Ki67 staining			0.367
40–69	8 ( 6)	2 ( 6)
70–79	8 ( 6)	0
80–89	12 ( 9)	1 ( 3)
≥90	18 (13)	7 (22)
Missing	89 (66)	22 (69)
Bcl6 staining			0.438
Positive	14 (10)	4 (13)
Negative	17 (13)	1 ( 3)
Not tested/Not reported	104 (76)	27 (84)
Bcl2 staining			0.655
Positive	27 (20)	9 (28)
Negative	2 ( 1)	0
Not tested/Not reported	106 (78)	23 (72)
CD20 staining			0.120
Positive	100 (74)	19 (59)
Negative	3 ( 2)	0
Not tested/Not reported	32 (24)	13 (41)
CD30 staining			0.469
Positive	34 (25)	9 (28)
Negative	5 ( 4)	0
Not tested/Not reported	96 (71)	23 (72)
Pathology			0.636
PTLD	91 (67)	19 (59)
DLBCL	39 (29)	12 (38)
Others	5 ( 4)	1 ( 3)
Subtype			0.963
DLBCL	58 (43)	15 (47)
Polymorphic PTLD	27 (20)	6 (19)
Burkitt	2 ( 1)	1 ( 3)
PTLD NOS	24 (18)	5 (16)
Plasmacytoid	13 (10)	3 ( 9)
T-cell NOS	2 ( 1)	1 ( 3)
Monomorphic PTLD	2 ( 1)	0
No morphology	4 ( 3)	0
Missing	3 ( 2)	1 ( 3)

Follow-up and survival

Median follow-up of survivors from initial alloHCT was 62 months (range, 3 – 167) for patients with EBV^pos PTLD and 49 months (range, 6 – 121) for EBV^neg PTLD. In univariate analysis, probability of OS following diagnosis of PTLD for the entire study cohort was 53% (47–59) at one year. Probability of OS following EBV^pos PTLD was 60% (53–66) at six months and 55% (48–61) at one year. The probability of OS following EBV^neg PTLD was 49% (34–63) at six months and 44% (29–58) at one year (Figure 1) and did not differ (p = 0.21) from patients with EBV^pos PTLD. Of the 154 patients who died, centers reported PTLD as a primary or contributory cause of death in 59 (38%) patients.

Figure 1: Overall Survival (OS) from diagnosis of PTLD for EBV-positive and EBV-negative PTLD.

PTLD: Post Transplant Lymphoproliferative disorder; OS: Overall Survival

In multivariable analysis (Table 3), survival did not differ based upon EBV status of PTLD (EBV^pos: HR 1.42, 95% CI 0.94–2.15, p=0.0976). Patients with malignant disease who received NMA/RIC conditioning with TBI had the highest risk for mortality (HR 2.06, 95% CI 1.11–3.82, p = 0.022). TCD with ATG and/or alemtuzumab resulted in poor clinical outcomes of PTLD (OS HR 2.09, 95% CI 1.34–3.26, p=0.0012).

Table 3.

Results of multivariable analysis for overall survival following diagnosis of PTLD

Variable	N	RR of death (95% Confidence Interval)	p-value
PTLD Type			0.098
EBV-positive	222	1.00
EBV-negative	45	1.42 (0.94 – 2.15)
Disease/Conditioning*			0.0002
Malignant disease/MAC+TBI	68	1.00
Malignant disease/MAC no TBI	59	1.42 (0.91 – 2.21	0.128
Malignant disease/RIC+TBI	18	2.06 (1.11 – 3.82)	0.022
Malignant disease/RIC no TBI	37	1.49 (0.91 – 2.44)	0.117
Non-malignant disease +TBI	35	0.42 (0.21 – 0.82)	0.011
Non-malignant disease no TBI	45	0.72 (0.42 – 1.23)	0.234
ATG/CAMPATH received			0.0012
No ATG/CAMPATH	56	1.00
ATG/CAMPATH given	206	2.09 (1.34 – 3.26)

TBI = total body irradiation; MAC: Myeloablative conditioning; NMA: Nonmyeloablative; RIC: Reduced intensity conditioning. ATG=anti-thymocyte globulin

Contrast analyses for the disease/conditioning variable combinations were performed. Only those combinations demonstrating an increased risk of death are reported here: Malignant disease/MAC no TBI vs Non-malignant disease + TBI [RR 3.41 (1.73 – 6.74); p< 0.001]; Malignant disease/MAC no TBI vs Non-malignant disease no TBI [RR 1.96 (1.15 – 3.35); p=0.014]; Malignant disease/RIC+TBI vs Non-malignant disease + TBI [RR 4.96 (2.23 –11.06), p< 0.001]; Malignant disease/RIC+TBI vs Non-malignant disease no TBI [RR 2.85 (1.44 – 5.65), p=0.003]; Malignant disease/RIC no TBI vs Non-malignant disease + TBI [RR 3.59 (1.78 – 7.23), p<0.001]; Malignant disease/RIC no TBI vs Non-malignant disease no TBI [RR 2.06 (1.17 – 3.63), p= 0.013].

DISCUSSION:

The current analysis reports the outcomes of 267 patients with EBV^pos and EBV^neg PTLD reported to the CIBMTR from 2002 to 2014. Overall, PTLD occurred in 1% of analyzable patients receiving alloHCT during the study period. To our knowledge, this is the largest report examining EBV^neg and EBV^pos PTLD in an alloHCT population. One year survival of patients who developed PTLD in this large cohort of alloHCT patients was 53% (47–59%). There was no difference in OS between EBV^pos and EBV^neg PTLD. Patients who developed PTLD following alloHCT for malignant disease and use of T-cell depletion (TCD) had inferior overall survival (OS). There was a non-significant trend toward later onset for EBV^neg PTLD compared to EBV^pos PTLD. Patients with EBV^neg PTLD had higher total WBC but a lower ALC at day 100 than patients with EBV^pos PTLD. Furthermore, for patients who developed CMV DNAemia preceding PTLD, the CMV DNAemia occurred later for EBV^neg PTLD patients.

In a previous CIBMTR analysis, Curtis et al reported on 18,014 patients from 235 centers worldwide in whom PTLD developed in 78 recipients (0.43%), with 82% occurring within the first year following alloHCT.⁸ In multivariable analyses, the risk of early-onset PTLD (<1 year) was significantly associated with unrelated or human leukocyte antigen (HLA)-mismatched related donor (RR=4.1, p<0.001), donor marrow TCD (RR=12.7, p<0.001), and use of ATG (RR= 6.4, p<0.001) or anti-CD3 monoclonal antibody (RR=43.2, p<0.001). In addition, PTLD associated with occurrence of grades II-IV aGvHD (RR=1.9, p=0.02) and with conditioning regimens that included radiation (RR=2.9, p=0.02). Extensive cGvHD was the sole risk factor for late onset PTLD (RR=4.0, p=0.01). Rates of PTLD among patients with 2 and ≥3 major risk factors were 8.0 ± 2.9% and 22 ± 17.9%, respectively. Lastly, use of ATG associated with higher incidence of PTLD compared to alemtuzumab (p=0.009). While the prior CIBMTR study focused on early- versus late-onset PTLD, the analysis did not include information on EBV^pos and EBV^neg PTLD. Interestingly, the median onset of PTLD for our patients was “early” by the Curtis et al definition as EBV^pos PTLD occurred at a median of 3 months and EBV^neg PTLD at a median of 5 months.

Styczynski and colleagues with the European Group for Blood and Marrow Transplantation (EBMT) reported outcomes of 86 proven and 58 probable PTLD following alloHCT, but also did not report on EBV^neg PTLD.¹⁷ Among 4,466 alloHCT, overall frequency of EBV related PTLD was 3.2% and varied based on donor type (MRD=1.16%, MMRD=2.86%, MUD=3.97%, MMUD=11.24% and UCB=4.06%). Late PTLD (>6 months after alloHCT) was reported in 11% (16/144). Investigators identified age at alloHCT (>30 years), extra nodal involvement, aGvHD greater than grade II, and absence of reduction in immunosuppression as poor prognostic factors. Probability of 3-year OS was 73%, 52%, and 26%, respectively, for patients with 0–1, 2 or 3 risk factors, respectively. Even though EBV^neg patients were not included in this report, overall incidence of PTLD at 3.2% is higher than our analysis and that previously reported by CIBMTR.¹³

TCD is a risk factor for PTLD, although the method of TCD may vary the PTLD incidence. TCD via ex vivo methodology appears associated with a high incidence of PTLD, occurring in 29% in a study of MUD alloHCT.^10,11 Few patients in the current CIBMTR study received ex vivo TCD (only 12% for EBV^pos and 4% for EBV^neg PTLD) precluding meaningful analysis. Similarly, in vivo TCD using ATG is also associated with a higher incidence of PTLD, reaching 21% in a report of 335 patients compared to an incidence of only 2% without ATG.¹¹ A marked increased risk of EBV-related complications including viremia and PTLD (21% vs 2%; p<0.01) were seen with the addition of ATG to a non-myeloablative conditioning prior to unrelated umbilical cord blood transplants.¹⁰ In contrast, in vivo TCD using alemtuzumab was associated with PTLD incidence of only 1.3% comparable to the incidence generally seen with unmanipulated graft alloHCT.¹² The difference between ATG and alemtuzumab is likely due to the latter’s depletion in both B and T cells. Alemtuzumab-associated, B-cell depletion possibly reduces the viral B-cell reservoir until T cells become fully functional following alloHCT. In the current study, patients receiving ATG or alemtuzumab had worse survival compared to those without in vivo TCD (HR 2.09, 95% CI 1.34–3.26, p=0.0012). The patients treated with alemtuzumab (n=18) were too few to separate from those receiving ATG. Interestingly, in vivo TCD was used similarly in both the EBV^pos and EBV^neg cohorts.

Other potential risks for PTLD may include issues related to alternative donors and alternative GVHD prophylaxis and donor/recipient EBV serostatus. For EBV serostatus, there may be a higher risk of developing PTLD in EBV positive recipients receiving cells from EBV seronegative donors since it is much harder to generate a primary immune response to EBV reactivation post alloHCT^3,18,19. For our cohort, EBV negative donors were used in 53 EBV positive recipients and 43 developed EBV^pos PTLD with the remainder developing EBV^neg PTLD. This study demonstrated varied incidence by donor and recipient HLA match; however due to few patients with haploidentical transplantation or receiving post-transplant cyclophosphamide no specific conclusions regarding risks in recipients of haploidentical donor or with post-transplant cyclophosphamide are possible²⁰.

There are notable differences between solid organ transplant (SOT) related and alloHCT-related PTLD and the incidence and outcomes of PTLD varies markedly when the reports contain both disease entities (Table 4)^5,21–24. Compared to SOT, PTLD following alloHCT develops earlier, has an aggressive course and poorer survival^22–24. In addition, SOT-associated PTLD has more frequent occurrence of EBV^neg disease⁷. Although there are several reports examining PTLD in alloHCT patients, very few describe EBV^neg PTLD. In the largest study, Fox et al described 74 patients that excluded histological EBV^neg PTLD and negative EBV DNAemia patients²⁵. Similarly, Hale et al (n=20) and Gerritsen, et al (n=65) provide no information of EBV positivity or absence of it ^12,26. Notably, our study found an incidence of 17% EBV^neg PTLD following alloHCT. This compares to a 33% incidence of EBV^neg PTLD in an analysis of 176 SOT patients²⁶. One factor proposed as a possible explanation for EBV^neg PTLD includes lymphoid stimulation from yet unidentified viral or other infections.^5,20 In spite of etiological factors, PTLD response to reduction in immunosuppression appears similar for EBV^pos and EBV^neg patients.

Table 4:

Summary of published data for solid organ and hematopoietic cell transplant patients with PTLD

Study Author/year of publication	Patients (n) Total (SOT/alloHCT)	Salient Features of EBVneg PTLD
Nelson 200023	17 (15/2)	- Advanced age compared to EBVpos PTLD (50y vs 40y) - Later PTLD onset compared to EBVpos PTLD (50 mo vs 10 mo) - Increased incidence in recent years - More often monomorphic PTLC - More aggressive course - Similar response to reducing immunosuppression
Reshef 2008 24	30 (30/0)	- Similar age compared to EBVpos PTLD - Late onset - More often monomorphic PTLD and DLBCL - Associated with failure of grafted organ - Similar response to reducing immunosuppression and Rituximab - Improved survival in renal transplant compared to lung transplant recipients
Luskin 201522	176 (176/0)	- Higher recent incidence of EBVneg PTLD - Late onset - More often monomorphic PTLD - Similar high risk features compared to EBVpos PTLD - Similar response to reducing immunosuppression and Rituximab - Similar overall survival compared to EBVpos PTLD
Leblond 19985	11(0/11)	- Later onset compared to EBVpos PTLD - More often monomorphic PTLD and DLBCL - Inferior survival compared to 21 matched patients without PTLD
Dotti 200021	8 (8/0)	- Later onset compared to EBVpos PTLD

The current study had limitations and strengths of a typical retrospective registry analysis. Our analysis includes 267 patients from 106 centers providing diffuse applicability. However, the nature of data captured in the CIBMTR registry precludes serial EBV viral load data monitoring or additional clinical indications for diagnostic work-up. Information regarding pre-emptive treatment with Rituximab or details regarding IVIG indication and timing are also lacking. Finally, the treatment and response following the PTLD diagnosis are unavailable. Information regarding immunosuppressive agents, including medications and doses, at the time of PTLD diagnosis are not readily available. Given the median onset of diagnosis was 3 months (EBV^pos) and 5 months (EBV^neg), it is presumed that the majority of these patients were significantly immunocompromised at PTLD onset. The other surrogate marker of on-going immunocompromised status is GVHD. We found that 31% of patients were diagnosed with grade II-IV acute GVHD and an additional 16% developed chronic GVHD preceding the PTLD diagnosis. However, our multivariable analysis did not find a diagnosis of GVHD prior to the onset of PTLD negatively influencing the survival of PTLD patients. This does not preclude GVHD as a risk factor for PTLD, only that GVHD preceding PTLD was not associated with survival following the diagnosis of PTLD. A central review of pathology reports for 167 (63%) patients was conducted. Although a review of actual pathology slides was not feasible, systematic review of the pathology reports for confirmation of reported EBV status of the PTLD was performed and fewer than 5% of cases were changed from the centers original report following pathology report review. Furthermore, 71% of EBV^neg PTLD cases were confirmed with pathology reports. For the 100 patients without pathology reports available, there were no differences in clinical features or outcomes compared to the pathologically reviewed cohort allowing for a large analysis including multivariable survival analysis. Our analysis sought to determine outcomes following the diagnosis but also examined risk factors present at transplant that impacted survival following the PTLD diagnosis. Determination of risk factors for development of PTLD were not examined due to lack of data on surveillance and center practice. While this may be achieved with a case-cohort analysis limiting a control population to the centers with cases, prior analyses have already identified PTLD risk factors. Our study instead focuses on potential differences in outcomes of EBV^pos and EBV^neg PTLD. The similar survival of EBV^pos and EBV^neg PTLD suggests that optimal treatment remains undefined.

In summary, we found that while there is no difference in survival outcomes for patients with EBV^pos or EBV^neg PTLD occurring following alloHCT, differences in WBC, ALC, time to onset, and frequency of preceding CMV DNAemia occurred. Furthermore, the survival is poor at 1 year following the PTLD diagnosis and is worse for patients who receive in vivo T-cell depletion or those who receive a TBI containing NMA/RIC conditioning for malignant disease. Further clinical studies are needed to define the incidence, biology and response to therapy for alloHCT-associated EBV^neg PTLD. The impact of novel therapies, including viral specific T-cells, may lead to significant improvement in outcomes for EBV^pos PTLD in the future.