Original Clinical Science
Expression of Intratumoral Forkhead Box Protein 3 in Posttransplant Lymphoproliferative Disorders: Clinical Features and Survival Outcomes David Berglund,1,2 Amelie Kinch,3 Elin Edman,4 Carin Backlin,5 Gunilla Enblad,6 Erik Larsson,2 Daniel Molin,6 Karlis Pauksens,3 Christer Sundström,2 and Eva Baecklund5 Background. The infiltration of regulatory T cells (Tregs) in lymphomas is associated with better prognosis for some types of lymphomas, but knowledge of their role in posttransplant lymphoproliferative disorders (PTLDs) is limited. We therefore investigated the association between the expression of the Treg marker forkhead box protein 3 (FoxP3) in biopsies of PTLDs and survival, PTLD subtype, and clinical characteristics. Methods. Seventy-four cases of PTLD after solid organ transplantation with sufficient material for further analysis were included from a population-based study of PTLDs in Sweden. The PTLD biopsies were reevaluated and stained with the 236A/E7 antibody to detect FoxP3 in lymphoma tissue. Detailed clinical data were collected retrospectively from medical records. Results. Based on a cutoff level of 29 FoxP3+ cells per mm2, most (80%) of the PTLDs were FoxP3−. Forty-seven of 74 PTLDs displayed no FoxP3+ cells at all. The frequency of FoxP3+ cells did not influence median overall survival. The FoxP3− PTLDs were more frequently of T-cell phenotype (P = 0.04), located at the graft (P = 0.03), occurred earlier after transplantation (P = 0.04), were more likely to develop in lung recipients (P = 0.04), and in patients that had received anti– T-cell globulin as induction therapy (P = 0.02). The FoxP3+ PTLDs were associated with hepatitis C seropositivity (P = 0.03). In multivariate analysis, B-cell PTLD and hepatitis C infection were independent predictors of FoxP3 positivity. Conclusion. Our findings suggest that intratumoral FoxP3+ Tregs do not influence survival in patients with PTLD. FoxP3+ Tregs are rare in PTLD, possibly because of heavy immunosuppression.
(Transplantation 2015;99: 1036–1042)
P
osttransplant lymphoproliferative disorders (PTLDs) are a well-recognized complication of solid organ transplantation and are associated with a significant morbidity and mortality rate.1,2 The development of PTLDs is mainly dependent on the use of immunosuppressive drugs and the loss of immunologic control of the Epstein-Barr virus (EBV).3‐5 The PTLD incidence is higher in nonrenal compared with renal transplant recipients because of the different immunosuppressive regimens.2 In particular, treatment with anti–T-cell antibodies, such as antithymocyte globulin (ATG) and OKT3, has been associated with an increased risk of PTLD.5‐7 Regulatory T cells (Tregs) are a subpopulation of T cells that have a role in preventing autoimmune disease, maintaining transplantation tolerance, and controlling tumor immunity.8
Natural Tregs are characterized by the expression of CD4, CD25, and the transcription factor forkhead box protein 3 (FoxP3). The FoxP3 is considered the most reliable phenotypic marker of natural Tregs, even though the expression of FoxP3 is not strictly confined to Tregs but can also be expressed by activated effector T cells.9 There are also other types of Tregs, including some of CD8 lineage, which were not investigated in this article.10 The upregulation of Tregs has been proposed as one mechanism by which tumor cells evade the immune system.11 Indeed, with few exceptions, high numbers of FoxP3+ cells in tumors have been reported to correlate with poor prognosis in epithelial cancers.12‐14 On the contrary, the presence of intratumoral Tregs has been reported to have a positive impact
Received 4 April 2014. Revision requested 25 April 2014.
D.B. has received research funding from Njurfonden and Professor Lars-Erik Gelin Memorial Foundation. A.K. has received research funding from the Lion’s Cancer Research Foundation, Selander’s Research Foundation, and Professor Lars-Erik Gelin Memorial Foundation. Göran Granath contributed with statistical advice and expertise.
Accepted 11 July 2014. 1 Department of Surgical Sciences, Section of Transplantation Surgery, Uppsala University, Uppsala, Sweden. 2 Department of Immunology, Genetics and Pathology, Uppsala University, Uppsala, Sweden. 3
Section of Infectious Diseases, Department of Medical Sciences, Uppsala University, Uppsala, Sweden. 4
Halmstad Hospital, Halmstad, Sweden.
5
Section of Rheumatology, Department of Medical Sciences, Uppsala University, Uppsala, Sweden. 6 Section of Oncology, Department of Radiology, Oncology and Radiation Sciences, Uppsala University, Uppsala, Sweden.
D.B., A.K., C.B., G.E., E.L., D.M., K.P., C.S., E.B. participated in research design. D.B., A.K., E.E., C.B., G.E., E.L., D.M., K.P., C.S., E.B. participated in writing of the article. D.B., A.K., E.E., C.B., G.E., E.L., D.M., K.P., C.S., and E.B. participated in performance of the research. D.B., A.K., E.E., G.E., E.L., D.M., K.P., C.S., E.B. participated in data analysis. Correspondence: Amelie Kinch, MD, Section of Infectious Diseases, Department of Medical Sciences, Uppsala University, S-751 85 Uppsala, Sweden. ([email protected] ) Copyright © 2014 Wolters Kluwer Health, Inc. All rights reserved.
The authors declare no conflicts of interest.
ISSN: 0041-1337/15/9905-1036
D.B. and A.K. shared first authorship.
DOI: 10.1097/TP.0000000000000415
1036
www.transplantjournal.com
Transplantation
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
■
May 2015
■
Volume 99
■
Number 5
Berglund et al
© 2015 Wolters Kluwer
on survival in several lymphomas, for example, Hodgkin lymphoma, follicular lymphoma, and some types of T-cell lymphoma.15‐19 The effect on diffuse large B-cell lymphoma (DLBCL) is not as clear.15,20,21 In B-cell malignancies, Tregs have a complex role, that is, they may protect the tumor cells from immune responses mediated by T cells as well as kill the tumor cells by antigen recognition on major histocompatibility complex class II molecules.22 Until now, there is only one study addressing the role of Tregs in PTLD which reported low levels of Tregs and no correlation to survival.23 However, this study was limited to 31 cases of monomorphic B-cell PTLD. If an increased frequency of Tregs correlates with poor prognosis, this would have an impact on the risk-benefit assessment for upcoming clinical trials with adoptive Treg therapy for the induction of transplantation tolerance and treatment of graft-versus-host disease.24,25 In fact, in the area of oncology Tregs are often regarded as an important and challenging immune escape mechanism. Thus, efforts are being directed at depleting Tregs in cancer immunotherapy.26 A larger study is needed to determine whether Tregs have a prognostic impact in PTLD. We hypothesized that intratumoral Tregs may affect survival in PTLD and that the frequency and impact of Tregs may be different depending on PTLD subtype and clinical characteristics. In the present study, we therefore correlated the occurrence of FoxP3+ cells in PTLD tissue to outcome, PTLD subtype, posttransplant immunosuppression, type of transplant, PTLD localization, and other clinical characteristics. RESULTS FoxP3 Expression in PTLD
Overall, in 47 of 74 (64%) PTLDs, no FoxP3+ cells were detected with the 236A/E7 antibody (Fig. 1A). Among the 27 cases in which FoxP3+ cells could be detected (Fig. 1B
1037
and C), the median level was 29 positive cells per mm2 (range, 2–123). Using this value as the cutoff, 20% of PTLDs was defined as FoxP3+ and 80% as FoxP3− (Table 1). We found no association between calendar period of PTLD diagnosis or laboratory where the block was originally embedded and FoxP3 expression in the biopsies (data not shown). The tonsil tissue used as a positive control showed high frequencies of FoxP3+ cells. To further rule out technical reasons for the findings of a low occurrence of FoxP3+ cells, stainings with another FoxP3 antibody were performed (10 cases stained with clone mAb22510) with concordant results (data not shown). FoxP3 Expression in PTLD Subtypes
The distribution of FoxP3+ cells differed among PTLD subtypes (Table 1). Half (3 of 6) of the polymorphic PTLDs, a quarter (12 of 52) of the monomorphic B-cell PTLDs, and none (0 of 14) of the monomorphic T-cell PTLDs were FoxP3+. Overall, B-cell PTLDs were more often FoxP3+ compared with T-cell PTLDs (P = 0.04). No difference was observed in the proportion of FoxP3+ cases between germinal center and nongerminal center subtypes of DLBCL (27% vs. 23%, P = 0.55) or EBV-encoded RNA (EBER)–positive and EBER-negative PTLDs (19% vs. 20%, P = 0.91). Association With Clinical Characteristics
The FoxP3+ PTLDs occurred later after transplantation (median, 6.7 vs. 3.9 years; P = 0.04) compared with FoxP3− PTLDs (Table 1). For type of transplant and localization of PTLD, all PTLDs in lung transplant recipients (n = 13) and all tumors localized in the allograft (n = 14) were FoxP3−. These findings were statistically significant when compared with all other organs combined (P = 0.04) and all other PTLD localizations (P = 0.03). All 74 patients were screened for hepatitis C pretransplant. Seven patients were seropositive for hepatitis C, of whom five
FIGURE 1. PTLD tissue with no FoxP3+ cells (A) and scattered FoxP3+ cells in brown (B). All individual PTLD cases (n = 27) with positive FoxP3 staining (C). The median among the positive cases was 29 FoxP3+ cells per mm2 (dashed line). PTLD, posttransplant lymphoproliferative disorder.
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
1038
Transplantation
■
May 2015
■
Volume 99
■
Number 5
www.transplantjournal.com
TABLE 1.
Associations between expression of FoxP3 in PTLD tissue, PTLD subtype, and clinical features All
FoxP3−
FoxP3+
P
n (%) or median years (range)
All PTLD subtype Polymorphic PTLD Monomorphic PTLD B-cell lymphomas DLBCL Germinal center type Nongerminal center type Aggressive B-cell lymphoma, unclassifiable B-cell lymphoma, unclassifiable T-cell lymphomasc Malignant lymphoma, unclassifiable EBER-positive PTLD Age at PTLD, years median (range) Sex, male Type of transplant Kidney (including one kidney+pancreas) Liver Heart Lung Time from transplantation to PTLD, years median (range) Localization of PTLD Nodal, only Extranodal Allograftf Bone marrow/blood CNS Liver Lung/pleura GI tract Immunosuppression ATG as induction therapy ATG as rejection therapy Corticosteroids Cyclosporine A Tacrolimus Azathioprine Mycophenolic acid Hepatitis C infection Autoimmune diseaseh
74
59 (80)
15 (20)
6 66 52 43 11 30 7 2 14 2 37 54 (7‐76) 41
3 (50) 54 (82) 40 (77) 32 (74) 8 (73) 23 (77) 6 (86) 2 14 (100) 2 30 (81) 52 (15‐76) 30 (73)
3 (50) 12 (18) 12 (23) 11 (26) 3 (27) 7 (23) 1 (14) 0 0 0 7 (19) 54 (7‐70) 11 (27)
43 7 11 13 4.6 (0.1‐24)
33 (77) 5 (71) 8 (73) 13 (100) 3.9 (0.1‐24)
10 (23) 2 (29) 3 (27) 0 6.7 (0.7‐19)
0.10a
0.55b
0.04d 0.91 0.63 0.10
0.04e 0.04
17 57 14 10 8 11 9 16
11 (65) 48 (84) 14 (100) 9 (90) 5 (63) 9 (82) 8 (89) 13 (81)
6 (35) 9 (16) 0 1 (10) 3 (38) 2 (18) 1 (11) 3 (19)
0.08
24 12 74 66 15 61 19 7 9
23 (40) 10 (17) 59 (100) 52 (88) 13 (22) 50 (85) 14 (24) 3 (5) 9 (15)
1 (8) 2 (15) 15 (100) 14 (93) 2 (13) 11 (73) 5 (33) 4 (27) 0
0.02 0.62 – 0.49 0.36 0.25 0.45 0.03 0.11
0.03g 0.35g 0.20g 0.59g 0.34g 0.59g
a
Polymorphic vs. monomorphic PTLD. Germinal center vs. nongerminal center type of DLBCL. Anaplastic large cell lymphoma, ALK- n = 4, Peripheral T-cell lymphoma, unspecified n = 3, Enteropathy-associated T-cell lymphoma n = 2, Hepatosplenic T-cell lymphoma n = 1, T lymphoblastic lymphoma n = 1, Aggressive T-cell lymphoma, unclassifiable n = 3. d T cell vs. B-cell monomorphic PTLD. e Recipients of lungs vs. other organs. f PTLD in allograft is only registered as allograft and not as a separate organ in this table. g Comparison with all other localizations. h Ulcerative colitis n = 2, in one case in combination with celiac disease, Crohn’s disease n = 2, one of them in combination with Sjögren’s syndrome, Sjo¨gren’s syndrome only n = 1, SLE n = 1, rheumatoid arthritis n = 1, juvenile arthritis n = 1, thyroiditis n = 1. FoxP3 positivity was defined as ≥29 positive cells per mm2. FoxP3, forkhead box P3; PTLD, posttransplant lymphoproliferative disorders; DLBCL, diffuse large B-cell lymphoma; EBER, EBV-encoded RNA; CNS, central nervous system; GI, gastrointestinal tract; ATG, antithymocyte globulin. b c
had a chronic infection confirmed by PCR analysis in plasma. Patients with chronic hepatitis C more frequently developed FoxP3+ PTLD (P = 0.03). No associations were found between FoxP3 expression and age at PTLD diagnosis, sex, presentation stage according
to Ann Arbor, risk assessment according to the age-adjusted International Prognostic Index, presence of EBV primary infection or cytomegalovirus infection, or episodes of acute or chronic rejections. All nine patients with an underlying autoimmune disease developed FoxP3− PTLD. The numbers,
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Berglund et al
© 2015 Wolters Kluwer
1039
clinical characteristics, and FoxP3 expression (Fig. 2). B-cell PTLD and hepatitis C seropositivity were correlated to a higher FoxP3 expression. Another multivariate analysis, the general discriminant analysis (GDA), revealed B-cell PTLD as a predictor of high expression of FoxP3 (P = 0.006), whereas hepatitis C seropositivity tended to predict high FoxP3 expression (P = 0.08). When only B-cell PTLD was included in the GDA, hepatitis C was independently associated with a high expression of FoxP3 (P = 0.049). Survival
FIGURE 2. Loading scatter plot from the partial least squares analysis for clinical features associated with FoxP3 expression in univariate analysis. FoxP3, forkhead box protein 3.
however, were too small to properly test the existence of an association between autoimmune disease and FoxP3− PTLD. All associations between FoxP3 expression and PTLD subtype or clinical characteristics that were found with the cutoff value of 29 positive cells per mm2 could also be confirmed with FoxP3 as a continuous variable. Association With Immunosuppressive Therapy
Thirty patients received ATG treatment (24 as induction therapy) (Table 1). Patients with FoxP3− PTLD had more often received induction therapy with ATG than patients with FoxP3+ PTLD (40% vs. 8%, P = 0.02). However, ATG as rejection therapy was equally common in the two groups. The median cumulative dose of ATG as induction therapy was higher in FoxP3− cases (P = 0.04), as was the median number of ATG doses (P = 0.03). There were no other differences in type of immunosuppressive therapy between the groups (dosage not compared). Multivariate Analysis of Factors Influencing FoxP3 Expression
The loading scatter plot from the partial least squares analysis shows the correlations between PTLD phenotype,
The median follow-up time was 9.2 years (range, 0.3–27) from transplantation until death or end of follow-up, which ever occurred first. At the end of follow-up (October 25, 2012), 63 patients had died (47 with FoxP3− and 16 with FoxP3+ PTLD). There was no difference in median overall survival between FoxP3+ and FoxP3− cases when using the cutoff value (P = 0.55, log-rank test, Fig. 3A), or when using FoxP3 as a continuous variable (P = 0.47, Cox proportional hazards regression). Further, there was no difference in median overall survival in the subgroup of monomorphic B-cell lymphomas, whether analyzing with the cutoff value (P = 0.72, Fig. 3B) or with FoxP3 as a continuous variable (P = 0.91). No differences were observed in survival based on the FoxP3 expression in the group of DLBCLs (P = 0.79, log-rank test), or in germinal center (P = 0.86, log-rank test) and nongerminal center (P = 0.99, log-rank test) subtypes of DLBCL. Relationship Between CD3 and FoxP3
Because activated T cells have been shown to transiently express FoxP3, we correlated the frequency of CD3+ cells to the number of FoxP3+ cells per mm2. No differences were observed in FoxP3 expression between biopsies with different frequencies of CD3+ cells (P = 0.99). DISCUSSION In this work, which is the largest study of Tregs in PTLD to date, we present 74 cases of PTLD that have been evaluated by FoxP3 staining. The majority of PTLDs displayed either no intratumoral FoxP3+ cells or a low frequency of these cells using the 236A/E7 antibody. This finding was also confirmed by the use of another antibody (mAb22510) in a sample of
FIGURE 3. Comparison of survival between FoxP3+ and FoxP3− cases of PTLD (A, P = 0.55) and of monomorphic B-cell PTLD (B, P = 0.72). PTLD, posttransplant lymphoproliferative disorder; FoxP3, forkhead box protein 3.
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
1040
Transplantation
■
May 2015
■
Volume 99
■
Number 5
cases. There is only one previous study on Tregs in PTLD tissue.23 This study by Richendollar et al. reported low levels of FoxP3+ cells in 30 cases of DLBCL and in one case of Burkitt lymphoma, but no completely negative cases using mAb22510. We observed lower numbers of FoxP3+ cells in B-cell PTLDs compared with Richendollar et al. and also compared with DLBCLs in the general population.15,20 Further, we found that B-cell PTLDs were more often FoxP3+ compared with T-cell PTLDs. No previous reports are available on intratumoral Tregs in T-cell PTLD. However, the presence of intratumoral Tregs has been reported in several types of T-cell lymphomas in the general population, including some subtypes in the present study, that is, anaplastic large cell lymphoma and peripheral T-cell lymphoma, unspecified.15 In contrast, of the 14 cases of T-cell PTLD in our study, 13 displayed no intratumoral FoxP3+ cells, and in the remaining case, there was a density of only 10 cells per mm2. Altogether, this suggests that Tregs are rare in PTLD and are more often found in B-cell PTLDs than T-cell PTLDs. The lower levels of intratumoral Tregs in PTLD compared with lymphomas in the general population may be because of the use of immunosuppressive drugs posttransplant. We found that patients receiving induction therapy with ATG more frequently developed a FoxP3− PTLD versus those not treated with ATG as induction. However, rabbit ATG has been reported to induce expansion of Tregs in vitro and in peripheral blood in kidney recipients, but there are no studies on how ATG influences the intratumoral immune response.27‐30 Evidence has emerged that calcineurin inhibitors decrease the level of FoxP3 in peripheral blood.31‐33 It might be that the FoxP3− PTLDs are more likely to occur in patients with heavier immunosuppression. Our findings that FoxP3− PTLD arose earlier after transplantation and was more frequently found in lung recipients support this hypothesis in that both these situations are associated with heavy immunosuppression.5 Hepatitis C seropositivity was independently associated with higher FoxP3 expression in B-cell PTLD. Higher concentrations of circulating Tregs have been reported in patients with chronic hepatitis C as compared with patients who have recovered from hepatitis C.34,35 These Tregs seem to play a role in viral persistence by suppressing hepatitis C–specific T-cell responses. Whether the higher levels of Tregs in hepatitis C patients influence the risk for or prognosis of PTLD remains to be investigated in another study. Epstein-Barr virus infection has been reported to promote the migration of Tregs to the microenvironment of the lymphoma.36 The EBV antigen EBNA-1, which is expressed in all EBV-positive lymphomas regardless of type of EBV latency, upregulates the expression of a chemokine in Hodgkin lymphoma cells, thereby stimulating the chemotaxis of Tregs. Further, in this study of Hodgkin lymphoma cells, Treg numbers were not observed to differ significantly with the EBV status of the lymphoma. The latter finding is in line with our study showing no association between presence of EBER and the expression of FoxP3 in PTLD tissue. As previously reported in monomorphic B-cell PTLDs,23 we confirm with a larger cohort and with several PTLD subtypes that overall survival of patients with PTLD is not affected by the frequency of intratumoral FoxP3. For DLBCLs in the general population, there have been conflicting results
www.transplantjournal.com
on the impact of infiltrating Tregs on outcome.15,20,21 It has been suggested that higher numbers of Tregs are associated with superior survival in germinal center DLBCL but inferior survival in nongerminal center DLBCL.15 In our study, FoxP3 expression did not influence survival in these subtypes of DLBCL, but the numbers were small. Further, FoxP3 expression in T-cell PTLD did not affect overall survival, a finding in accordance with studies on the same subtypes of T-cell lymphomas in the general population.15 Adoptive Treg cell therapy is a new therapeutic option in transplantation with the aim to decrease the need for general immunosuppression and possibly promote transplant tolerance.24,25 The potential risks with Treg immunotherapy include increased susceptibility to infections and cancers.25 Tregs have, in fact, been proposed as an important immune escape mechanism for malignant tumors.37 In the present study, we show that the expression of intratumoral FoxP3 in PTLD after solid organ transplantation is low (most likely because of the continuous administration of immunosuppressive drugs) and does not influence survival. These results do not support a central role for Tregs in PTLD and therefore indicate a favorable risk-benefit assessment in future Treg clinical trials. The strengths of this study are the large number of reevaluated PTLD cases, manually counted FoxP3+ cells, and detailed clinical data. Comparison with other studies may be dependent on two factors: the FoxP3 antibody clone used and the studied population. We opted to use the 236A/E7 antibody because it has been used previously to detect FoxP3 in solid tumors,14 Hodgkin lymphoma,17,38 and DLBCL,21 where it was shown to be predictive of the outcome. At present, we also use this antibody at our transplant center for routine screening of FoxP3+ cells in kidney transplant biopsies and duodenal biopsies of pancreas transplants. Because we observed large numbers of FoxP3+ cells in positive controls, we have no reason to suspect low sensitivity of the antibody staining. That we found no correlation between the density of CD3+ cells and FoxP3+ cells in the PTLD tissue further demonstrates that the 236A/E7 antibody is also specific for Tregs and does not detect T cells in general. The limitations of this study include the small subgroups for some of the PTLD subtypes and several of the clinical characteristics. Further, determining the threshold for FoxP3 positivity was not unambiguous. We decided to use the median frequency among the cases in which FoxP3+ cells were detected (29 FoxP3+ cells per mm2). An alternative would have been to define a case as positive based only on one positive cell per mm2; however, this is generally not an accepted practice in clinical pathology. Previous studies have typically used the level that gives improved survival.15,21 In our study, expression of FoxP3 did not influence survival and hence we could not choose a threshold using this method. Importantly, both the analyses of survival and of the association with clinical features gave the same results, regardless of the strategy to identify positive cases (i.e., 29 FoxP3+ cells per mm2 as the threshold versus using the frequency as a continuous variable). This finding supports the argument that defining a threshold for diagnostic purposes does not necessarily result in skewing of the data. In conclusion, the frequency of FoxP3+ cells in PTLDs, including posttransplant DLBCLs and T-cell lymphomas, was lower than previously reported for DLBCLs and T-cell
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Berglund et al
© 2015 Wolters Kluwer
lymphomas in immunocompetent individuals. Most importantly, intratumoral FoxP3+ cells did not affect survival. Indirectly, these findings do not support the commonly held concern that adoptive Treg therapy may increase the risk for PTLD. MATERIALS AND METHODS Patients, PTLD Diagnosis, and EBV Status
From a previously reported population-based cohort of 135 PTLDs after solid organ transplantation gathered in Sweden between 1980 and 2006,39 all cases of DLBCL, T-cell lymphoma, polymorphic PTLD, and PTLD localized to the graft or the lung with sufficient tissue available for further analysis were included in this study (n = 74). The tumor biopsies have been reevaluated according to the 2008 World Health Organization classification of lymphomas by an experienced hematopathologist (C.S.).40 The DLBCLs were classified as either germinal center or nongerminal center subtype using the Hans algorithm.41 Presence of EBV in PTLD tissue was analyzed by EBER in situ hybridization in 72 of the 74 cases. Detailed clinical data were retrieved retrospectively from medical records at the transplantation centers as well as from the hospitals where the patients were followed after transplantation. Diagnosis of EBV, cytomegalovirus, and hepatitis C was performed with the methods available at that time and vary over the decades. An assessment was made for every patient based on clinical symptoms, rise in antibody titres, antigen tests, and PCR assays. Autoimmune diseases were defined as all autoimmune diseases noted in the medical records and present before transplantation. Diabetes mellitus type 1 was not regarded as an autoimmune disease in this study. The study was approved by the Regional Ethical Review Board in Uppsala, Sweden. Immunohistochemistry for FoxP3
Tissue sections (4 μm thick) were made of formalin-fixed, paraffin-embedded diagnostic biopsies of PTLD. Deparaffination and antigen retrieval (Tris-ethylenediaminetetraacetic acid buffer, pH 9) were performed with a PT-link (Dako, Glostrup, Denmark). A Dako autostainer plus was used for immunohistochemical staining, including the mouse anti-human FoxP3 antibody (dilution 1:100, clone 236A/E7; eBioscience, San Diego, CA) and the EnVision DAB kit (Dako). Ten cases were stained with another mouse anti-human FoxP3 antibody (clone mAb22510; Abcam, Cambridge, MA) for comparison. Formalin-fixed, paraffin-embedded tissue sections of human tonsils were used as positive controls and stained in the same way as the PTLD tissue. The tonsil tissues were obtained from otherwise healthy patients undergoing routine tonsillectomy. Slides already stained for CD3 (n = 45) were retrieved from local pathology laboratories to estimate the frequency of CD3+ cells in each PTLD specimen. Cell Counting and Definition of FoxP3+ PTLD
Stained PTLD biopsies were assessed for FoxP3+ cells by manual counting using a light microscope (Leica DMD 108, Leica Microsystems Inc.) with the support of a grid. To create a cell-to-area ratio (expressed as of FoxP3+ cells per mm2), the area of each biopsy was calculated using the preinstalled software of the microscope. All biopsies were analyzed in their entirety, except for necrotic areas that
1041
were excluded. The analysed area of the biopsies with detectable FoxP3+ cells ranged from 1.7 to 176 mm2 (median, 44 mm2). FoxP3+ PTLD was defined as a case with 29 or more positive cells per mm2, which was the median density among the cases in which FoxP3 could be detected (Fig. 1C). The frequency of CD3+ cells was categorized into four groups based on their proportion of lymphocytes in the biopsy: 0% to less than 25%, 25% to less than 50%, 50% to less than 75%, and 75% to 100%. Statistics
The chi-square test or Fisher exact test (if
Expression of Intratumoral Forkhead Box Protein 3 in Posttransplant Lymphoproliferative Disorders: Clinical Features and Survival Outcomes David Berglund,1,2 Amelie Kinch,3 Elin Edman,4 Carin Backlin,5 Gunilla Enblad,6 Erik Larsson,2 Daniel Molin,6 Karlis Pauksens,3 Christer Sundström,2 and Eva Baecklund5 Background. The infiltration of regulatory T cells (Tregs) in lymphomas is associated with better prognosis for some types of lymphomas, but knowledge of their role in posttransplant lymphoproliferative disorders (PTLDs) is limited. We therefore investigated the association between the expression of the Treg marker forkhead box protein 3 (FoxP3) in biopsies of PTLDs and survival, PTLD subtype, and clinical characteristics. Methods. Seventy-four cases of PTLD after solid organ transplantation with sufficient material for further analysis were included from a population-based study of PTLDs in Sweden. The PTLD biopsies were reevaluated and stained with the 236A/E7 antibody to detect FoxP3 in lymphoma tissue. Detailed clinical data were collected retrospectively from medical records. Results. Based on a cutoff level of 29 FoxP3+ cells per mm2, most (80%) of the PTLDs were FoxP3−. Forty-seven of 74 PTLDs displayed no FoxP3+ cells at all. The frequency of FoxP3+ cells did not influence median overall survival. The FoxP3− PTLDs were more frequently of T-cell phenotype (P = 0.04), located at the graft (P = 0.03), occurred earlier after transplantation (P = 0.04), were more likely to develop in lung recipients (P = 0.04), and in patients that had received anti– T-cell globulin as induction therapy (P = 0.02). The FoxP3+ PTLDs were associated with hepatitis C seropositivity (P = 0.03). In multivariate analysis, B-cell PTLD and hepatitis C infection were independent predictors of FoxP3 positivity. Conclusion. Our findings suggest that intratumoral FoxP3+ Tregs do not influence survival in patients with PTLD. FoxP3+ Tregs are rare in PTLD, possibly because of heavy immunosuppression.
(Transplantation 2015;99: 1036–1042)
P
osttransplant lymphoproliferative disorders (PTLDs) are a well-recognized complication of solid organ transplantation and are associated with a significant morbidity and mortality rate.1,2 The development of PTLDs is mainly dependent on the use of immunosuppressive drugs and the loss of immunologic control of the Epstein-Barr virus (EBV).3‐5 The PTLD incidence is higher in nonrenal compared with renal transplant recipients because of the different immunosuppressive regimens.2 In particular, treatment with anti–T-cell antibodies, such as antithymocyte globulin (ATG) and OKT3, has been associated with an increased risk of PTLD.5‐7 Regulatory T cells (Tregs) are a subpopulation of T cells that have a role in preventing autoimmune disease, maintaining transplantation tolerance, and controlling tumor immunity.8
Natural Tregs are characterized by the expression of CD4, CD25, and the transcription factor forkhead box protein 3 (FoxP3). The FoxP3 is considered the most reliable phenotypic marker of natural Tregs, even though the expression of FoxP3 is not strictly confined to Tregs but can also be expressed by activated effector T cells.9 There are also other types of Tregs, including some of CD8 lineage, which were not investigated in this article.10 The upregulation of Tregs has been proposed as one mechanism by which tumor cells evade the immune system.11 Indeed, with few exceptions, high numbers of FoxP3+ cells in tumors have been reported to correlate with poor prognosis in epithelial cancers.12‐14 On the contrary, the presence of intratumoral Tregs has been reported to have a positive impact
Received 4 April 2014. Revision requested 25 April 2014.
D.B. has received research funding from Njurfonden and Professor Lars-Erik Gelin Memorial Foundation. A.K. has received research funding from the Lion’s Cancer Research Foundation, Selander’s Research Foundation, and Professor Lars-Erik Gelin Memorial Foundation. Göran Granath contributed with statistical advice and expertise.
Accepted 11 July 2014. 1 Department of Surgical Sciences, Section of Transplantation Surgery, Uppsala University, Uppsala, Sweden. 2 Department of Immunology, Genetics and Pathology, Uppsala University, Uppsala, Sweden. 3
Section of Infectious Diseases, Department of Medical Sciences, Uppsala University, Uppsala, Sweden. 4
Halmstad Hospital, Halmstad, Sweden.
5
Section of Rheumatology, Department of Medical Sciences, Uppsala University, Uppsala, Sweden. 6 Section of Oncology, Department of Radiology, Oncology and Radiation Sciences, Uppsala University, Uppsala, Sweden.
D.B., A.K., C.B., G.E., E.L., D.M., K.P., C.S., E.B. participated in research design. D.B., A.K., E.E., C.B., G.E., E.L., D.M., K.P., C.S., E.B. participated in writing of the article. D.B., A.K., E.E., C.B., G.E., E.L., D.M., K.P., C.S., and E.B. participated in performance of the research. D.B., A.K., E.E., G.E., E.L., D.M., K.P., C.S., E.B. participated in data analysis. Correspondence: Amelie Kinch, MD, Section of Infectious Diseases, Department of Medical Sciences, Uppsala University, S-751 85 Uppsala, Sweden. ([email protected] ) Copyright © 2014 Wolters Kluwer Health, Inc. All rights reserved.
The authors declare no conflicts of interest.
ISSN: 0041-1337/15/9905-1036
D.B. and A.K. shared first authorship.
DOI: 10.1097/TP.0000000000000415
1036
www.transplantjournal.com
Transplantation
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
■
May 2015
■
Volume 99
■
Number 5
Berglund et al
© 2015 Wolters Kluwer
on survival in several lymphomas, for example, Hodgkin lymphoma, follicular lymphoma, and some types of T-cell lymphoma.15‐19 The effect on diffuse large B-cell lymphoma (DLBCL) is not as clear.15,20,21 In B-cell malignancies, Tregs have a complex role, that is, they may protect the tumor cells from immune responses mediated by T cells as well as kill the tumor cells by antigen recognition on major histocompatibility complex class II molecules.22 Until now, there is only one study addressing the role of Tregs in PTLD which reported low levels of Tregs and no correlation to survival.23 However, this study was limited to 31 cases of monomorphic B-cell PTLD. If an increased frequency of Tregs correlates with poor prognosis, this would have an impact on the risk-benefit assessment for upcoming clinical trials with adoptive Treg therapy for the induction of transplantation tolerance and treatment of graft-versus-host disease.24,25 In fact, in the area of oncology Tregs are often regarded as an important and challenging immune escape mechanism. Thus, efforts are being directed at depleting Tregs in cancer immunotherapy.26 A larger study is needed to determine whether Tregs have a prognostic impact in PTLD. We hypothesized that intratumoral Tregs may affect survival in PTLD and that the frequency and impact of Tregs may be different depending on PTLD subtype and clinical characteristics. In the present study, we therefore correlated the occurrence of FoxP3+ cells in PTLD tissue to outcome, PTLD subtype, posttransplant immunosuppression, type of transplant, PTLD localization, and other clinical characteristics. RESULTS FoxP3 Expression in PTLD
Overall, in 47 of 74 (64%) PTLDs, no FoxP3+ cells were detected with the 236A/E7 antibody (Fig. 1A). Among the 27 cases in which FoxP3+ cells could be detected (Fig. 1B
1037
and C), the median level was 29 positive cells per mm2 (range, 2–123). Using this value as the cutoff, 20% of PTLDs was defined as FoxP3+ and 80% as FoxP3− (Table 1). We found no association between calendar period of PTLD diagnosis or laboratory where the block was originally embedded and FoxP3 expression in the biopsies (data not shown). The tonsil tissue used as a positive control showed high frequencies of FoxP3+ cells. To further rule out technical reasons for the findings of a low occurrence of FoxP3+ cells, stainings with another FoxP3 antibody were performed (10 cases stained with clone mAb22510) with concordant results (data not shown). FoxP3 Expression in PTLD Subtypes
The distribution of FoxP3+ cells differed among PTLD subtypes (Table 1). Half (3 of 6) of the polymorphic PTLDs, a quarter (12 of 52) of the monomorphic B-cell PTLDs, and none (0 of 14) of the monomorphic T-cell PTLDs were FoxP3+. Overall, B-cell PTLDs were more often FoxP3+ compared with T-cell PTLDs (P = 0.04). No difference was observed in the proportion of FoxP3+ cases between germinal center and nongerminal center subtypes of DLBCL (27% vs. 23%, P = 0.55) or EBV-encoded RNA (EBER)–positive and EBER-negative PTLDs (19% vs. 20%, P = 0.91). Association With Clinical Characteristics
The FoxP3+ PTLDs occurred later after transplantation (median, 6.7 vs. 3.9 years; P = 0.04) compared with FoxP3− PTLDs (Table 1). For type of transplant and localization of PTLD, all PTLDs in lung transplant recipients (n = 13) and all tumors localized in the allograft (n = 14) were FoxP3−. These findings were statistically significant when compared with all other organs combined (P = 0.04) and all other PTLD localizations (P = 0.03). All 74 patients were screened for hepatitis C pretransplant. Seven patients were seropositive for hepatitis C, of whom five
FIGURE 1. PTLD tissue with no FoxP3+ cells (A) and scattered FoxP3+ cells in brown (B). All individual PTLD cases (n = 27) with positive FoxP3 staining (C). The median among the positive cases was 29 FoxP3+ cells per mm2 (dashed line). PTLD, posttransplant lymphoproliferative disorder.
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
1038
Transplantation
■
May 2015
■
Volume 99
■
Number 5
www.transplantjournal.com
TABLE 1.
Associations between expression of FoxP3 in PTLD tissue, PTLD subtype, and clinical features All
FoxP3−
FoxP3+
P
n (%) or median years (range)
All PTLD subtype Polymorphic PTLD Monomorphic PTLD B-cell lymphomas DLBCL Germinal center type Nongerminal center type Aggressive B-cell lymphoma, unclassifiable B-cell lymphoma, unclassifiable T-cell lymphomasc Malignant lymphoma, unclassifiable EBER-positive PTLD Age at PTLD, years median (range) Sex, male Type of transplant Kidney (including one kidney+pancreas) Liver Heart Lung Time from transplantation to PTLD, years median (range) Localization of PTLD Nodal, only Extranodal Allograftf Bone marrow/blood CNS Liver Lung/pleura GI tract Immunosuppression ATG as induction therapy ATG as rejection therapy Corticosteroids Cyclosporine A Tacrolimus Azathioprine Mycophenolic acid Hepatitis C infection Autoimmune diseaseh
74
59 (80)
15 (20)
6 66 52 43 11 30 7 2 14 2 37 54 (7‐76) 41
3 (50) 54 (82) 40 (77) 32 (74) 8 (73) 23 (77) 6 (86) 2 14 (100) 2 30 (81) 52 (15‐76) 30 (73)
3 (50) 12 (18) 12 (23) 11 (26) 3 (27) 7 (23) 1 (14) 0 0 0 7 (19) 54 (7‐70) 11 (27)
43 7 11 13 4.6 (0.1‐24)
33 (77) 5 (71) 8 (73) 13 (100) 3.9 (0.1‐24)
10 (23) 2 (29) 3 (27) 0 6.7 (0.7‐19)
0.10a
0.55b
0.04d 0.91 0.63 0.10
0.04e 0.04
17 57 14 10 8 11 9 16
11 (65) 48 (84) 14 (100) 9 (90) 5 (63) 9 (82) 8 (89) 13 (81)
6 (35) 9 (16) 0 1 (10) 3 (38) 2 (18) 1 (11) 3 (19)
0.08
24 12 74 66 15 61 19 7 9
23 (40) 10 (17) 59 (100) 52 (88) 13 (22) 50 (85) 14 (24) 3 (5) 9 (15)
1 (8) 2 (15) 15 (100) 14 (93) 2 (13) 11 (73) 5 (33) 4 (27) 0
0.02 0.62 – 0.49 0.36 0.25 0.45 0.03 0.11
0.03g 0.35g 0.20g 0.59g 0.34g 0.59g
a
Polymorphic vs. monomorphic PTLD. Germinal center vs. nongerminal center type of DLBCL. Anaplastic large cell lymphoma, ALK- n = 4, Peripheral T-cell lymphoma, unspecified n = 3, Enteropathy-associated T-cell lymphoma n = 2, Hepatosplenic T-cell lymphoma n = 1, T lymphoblastic lymphoma n = 1, Aggressive T-cell lymphoma, unclassifiable n = 3. d T cell vs. B-cell monomorphic PTLD. e Recipients of lungs vs. other organs. f PTLD in allograft is only registered as allograft and not as a separate organ in this table. g Comparison with all other localizations. h Ulcerative colitis n = 2, in one case in combination with celiac disease, Crohn’s disease n = 2, one of them in combination with Sjögren’s syndrome, Sjo¨gren’s syndrome only n = 1, SLE n = 1, rheumatoid arthritis n = 1, juvenile arthritis n = 1, thyroiditis n = 1. FoxP3 positivity was defined as ≥29 positive cells per mm2. FoxP3, forkhead box P3; PTLD, posttransplant lymphoproliferative disorders; DLBCL, diffuse large B-cell lymphoma; EBER, EBV-encoded RNA; CNS, central nervous system; GI, gastrointestinal tract; ATG, antithymocyte globulin. b c
had a chronic infection confirmed by PCR analysis in plasma. Patients with chronic hepatitis C more frequently developed FoxP3+ PTLD (P = 0.03). No associations were found between FoxP3 expression and age at PTLD diagnosis, sex, presentation stage according
to Ann Arbor, risk assessment according to the age-adjusted International Prognostic Index, presence of EBV primary infection or cytomegalovirus infection, or episodes of acute or chronic rejections. All nine patients with an underlying autoimmune disease developed FoxP3− PTLD. The numbers,
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Berglund et al
© 2015 Wolters Kluwer
1039
clinical characteristics, and FoxP3 expression (Fig. 2). B-cell PTLD and hepatitis C seropositivity were correlated to a higher FoxP3 expression. Another multivariate analysis, the general discriminant analysis (GDA), revealed B-cell PTLD as a predictor of high expression of FoxP3 (P = 0.006), whereas hepatitis C seropositivity tended to predict high FoxP3 expression (P = 0.08). When only B-cell PTLD was included in the GDA, hepatitis C was independently associated with a high expression of FoxP3 (P = 0.049). Survival
FIGURE 2. Loading scatter plot from the partial least squares analysis for clinical features associated with FoxP3 expression in univariate analysis. FoxP3, forkhead box protein 3.
however, were too small to properly test the existence of an association between autoimmune disease and FoxP3− PTLD. All associations between FoxP3 expression and PTLD subtype or clinical characteristics that were found with the cutoff value of 29 positive cells per mm2 could also be confirmed with FoxP3 as a continuous variable. Association With Immunosuppressive Therapy
Thirty patients received ATG treatment (24 as induction therapy) (Table 1). Patients with FoxP3− PTLD had more often received induction therapy with ATG than patients with FoxP3+ PTLD (40% vs. 8%, P = 0.02). However, ATG as rejection therapy was equally common in the two groups. The median cumulative dose of ATG as induction therapy was higher in FoxP3− cases (P = 0.04), as was the median number of ATG doses (P = 0.03). There were no other differences in type of immunosuppressive therapy between the groups (dosage not compared). Multivariate Analysis of Factors Influencing FoxP3 Expression
The loading scatter plot from the partial least squares analysis shows the correlations between PTLD phenotype,
The median follow-up time was 9.2 years (range, 0.3–27) from transplantation until death or end of follow-up, which ever occurred first. At the end of follow-up (October 25, 2012), 63 patients had died (47 with FoxP3− and 16 with FoxP3+ PTLD). There was no difference in median overall survival between FoxP3+ and FoxP3− cases when using the cutoff value (P = 0.55, log-rank test, Fig. 3A), or when using FoxP3 as a continuous variable (P = 0.47, Cox proportional hazards regression). Further, there was no difference in median overall survival in the subgroup of monomorphic B-cell lymphomas, whether analyzing with the cutoff value (P = 0.72, Fig. 3B) or with FoxP3 as a continuous variable (P = 0.91). No differences were observed in survival based on the FoxP3 expression in the group of DLBCLs (P = 0.79, log-rank test), or in germinal center (P = 0.86, log-rank test) and nongerminal center (P = 0.99, log-rank test) subtypes of DLBCL. Relationship Between CD3 and FoxP3
Because activated T cells have been shown to transiently express FoxP3, we correlated the frequency of CD3+ cells to the number of FoxP3+ cells per mm2. No differences were observed in FoxP3 expression between biopsies with different frequencies of CD3+ cells (P = 0.99). DISCUSSION In this work, which is the largest study of Tregs in PTLD to date, we present 74 cases of PTLD that have been evaluated by FoxP3 staining. The majority of PTLDs displayed either no intratumoral FoxP3+ cells or a low frequency of these cells using the 236A/E7 antibody. This finding was also confirmed by the use of another antibody (mAb22510) in a sample of
FIGURE 3. Comparison of survival between FoxP3+ and FoxP3− cases of PTLD (A, P = 0.55) and of monomorphic B-cell PTLD (B, P = 0.72). PTLD, posttransplant lymphoproliferative disorder; FoxP3, forkhead box protein 3.
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
1040
Transplantation
■
May 2015
■
Volume 99
■
Number 5
cases. There is only one previous study on Tregs in PTLD tissue.23 This study by Richendollar et al. reported low levels of FoxP3+ cells in 30 cases of DLBCL and in one case of Burkitt lymphoma, but no completely negative cases using mAb22510. We observed lower numbers of FoxP3+ cells in B-cell PTLDs compared with Richendollar et al. and also compared with DLBCLs in the general population.15,20 Further, we found that B-cell PTLDs were more often FoxP3+ compared with T-cell PTLDs. No previous reports are available on intratumoral Tregs in T-cell PTLD. However, the presence of intratumoral Tregs has been reported in several types of T-cell lymphomas in the general population, including some subtypes in the present study, that is, anaplastic large cell lymphoma and peripheral T-cell lymphoma, unspecified.15 In contrast, of the 14 cases of T-cell PTLD in our study, 13 displayed no intratumoral FoxP3+ cells, and in the remaining case, there was a density of only 10 cells per mm2. Altogether, this suggests that Tregs are rare in PTLD and are more often found in B-cell PTLDs than T-cell PTLDs. The lower levels of intratumoral Tregs in PTLD compared with lymphomas in the general population may be because of the use of immunosuppressive drugs posttransplant. We found that patients receiving induction therapy with ATG more frequently developed a FoxP3− PTLD versus those not treated with ATG as induction. However, rabbit ATG has been reported to induce expansion of Tregs in vitro and in peripheral blood in kidney recipients, but there are no studies on how ATG influences the intratumoral immune response.27‐30 Evidence has emerged that calcineurin inhibitors decrease the level of FoxP3 in peripheral blood.31‐33 It might be that the FoxP3− PTLDs are more likely to occur in patients with heavier immunosuppression. Our findings that FoxP3− PTLD arose earlier after transplantation and was more frequently found in lung recipients support this hypothesis in that both these situations are associated with heavy immunosuppression.5 Hepatitis C seropositivity was independently associated with higher FoxP3 expression in B-cell PTLD. Higher concentrations of circulating Tregs have been reported in patients with chronic hepatitis C as compared with patients who have recovered from hepatitis C.34,35 These Tregs seem to play a role in viral persistence by suppressing hepatitis C–specific T-cell responses. Whether the higher levels of Tregs in hepatitis C patients influence the risk for or prognosis of PTLD remains to be investigated in another study. Epstein-Barr virus infection has been reported to promote the migration of Tregs to the microenvironment of the lymphoma.36 The EBV antigen EBNA-1, which is expressed in all EBV-positive lymphomas regardless of type of EBV latency, upregulates the expression of a chemokine in Hodgkin lymphoma cells, thereby stimulating the chemotaxis of Tregs. Further, in this study of Hodgkin lymphoma cells, Treg numbers were not observed to differ significantly with the EBV status of the lymphoma. The latter finding is in line with our study showing no association between presence of EBER and the expression of FoxP3 in PTLD tissue. As previously reported in monomorphic B-cell PTLDs,23 we confirm with a larger cohort and with several PTLD subtypes that overall survival of patients with PTLD is not affected by the frequency of intratumoral FoxP3. For DLBCLs in the general population, there have been conflicting results
www.transplantjournal.com
on the impact of infiltrating Tregs on outcome.15,20,21 It has been suggested that higher numbers of Tregs are associated with superior survival in germinal center DLBCL but inferior survival in nongerminal center DLBCL.15 In our study, FoxP3 expression did not influence survival in these subtypes of DLBCL, but the numbers were small. Further, FoxP3 expression in T-cell PTLD did not affect overall survival, a finding in accordance with studies on the same subtypes of T-cell lymphomas in the general population.15 Adoptive Treg cell therapy is a new therapeutic option in transplantation with the aim to decrease the need for general immunosuppression and possibly promote transplant tolerance.24,25 The potential risks with Treg immunotherapy include increased susceptibility to infections and cancers.25 Tregs have, in fact, been proposed as an important immune escape mechanism for malignant tumors.37 In the present study, we show that the expression of intratumoral FoxP3 in PTLD after solid organ transplantation is low (most likely because of the continuous administration of immunosuppressive drugs) and does not influence survival. These results do not support a central role for Tregs in PTLD and therefore indicate a favorable risk-benefit assessment in future Treg clinical trials. The strengths of this study are the large number of reevaluated PTLD cases, manually counted FoxP3+ cells, and detailed clinical data. Comparison with other studies may be dependent on two factors: the FoxP3 antibody clone used and the studied population. We opted to use the 236A/E7 antibody because it has been used previously to detect FoxP3 in solid tumors,14 Hodgkin lymphoma,17,38 and DLBCL,21 where it was shown to be predictive of the outcome. At present, we also use this antibody at our transplant center for routine screening of FoxP3+ cells in kidney transplant biopsies and duodenal biopsies of pancreas transplants. Because we observed large numbers of FoxP3+ cells in positive controls, we have no reason to suspect low sensitivity of the antibody staining. That we found no correlation between the density of CD3+ cells and FoxP3+ cells in the PTLD tissue further demonstrates that the 236A/E7 antibody is also specific for Tregs and does not detect T cells in general. The limitations of this study include the small subgroups for some of the PTLD subtypes and several of the clinical characteristics. Further, determining the threshold for FoxP3 positivity was not unambiguous. We decided to use the median frequency among the cases in which FoxP3+ cells were detected (29 FoxP3+ cells per mm2). An alternative would have been to define a case as positive based only on one positive cell per mm2; however, this is generally not an accepted practice in clinical pathology. Previous studies have typically used the level that gives improved survival.15,21 In our study, expression of FoxP3 did not influence survival and hence we could not choose a threshold using this method. Importantly, both the analyses of survival and of the association with clinical features gave the same results, regardless of the strategy to identify positive cases (i.e., 29 FoxP3+ cells per mm2 as the threshold versus using the frequency as a continuous variable). This finding supports the argument that defining a threshold for diagnostic purposes does not necessarily result in skewing of the data. In conclusion, the frequency of FoxP3+ cells in PTLDs, including posttransplant DLBCLs and T-cell lymphomas, was lower than previously reported for DLBCLs and T-cell
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Berglund et al
© 2015 Wolters Kluwer
lymphomas in immunocompetent individuals. Most importantly, intratumoral FoxP3+ cells did not affect survival. Indirectly, these findings do not support the commonly held concern that adoptive Treg therapy may increase the risk for PTLD. MATERIALS AND METHODS Patients, PTLD Diagnosis, and EBV Status
From a previously reported population-based cohort of 135 PTLDs after solid organ transplantation gathered in Sweden between 1980 and 2006,39 all cases of DLBCL, T-cell lymphoma, polymorphic PTLD, and PTLD localized to the graft or the lung with sufficient tissue available for further analysis were included in this study (n = 74). The tumor biopsies have been reevaluated according to the 2008 World Health Organization classification of lymphomas by an experienced hematopathologist (C.S.).40 The DLBCLs were classified as either germinal center or nongerminal center subtype using the Hans algorithm.41 Presence of EBV in PTLD tissue was analyzed by EBER in situ hybridization in 72 of the 74 cases. Detailed clinical data were retrieved retrospectively from medical records at the transplantation centers as well as from the hospitals where the patients were followed after transplantation. Diagnosis of EBV, cytomegalovirus, and hepatitis C was performed with the methods available at that time and vary over the decades. An assessment was made for every patient based on clinical symptoms, rise in antibody titres, antigen tests, and PCR assays. Autoimmune diseases were defined as all autoimmune diseases noted in the medical records and present before transplantation. Diabetes mellitus type 1 was not regarded as an autoimmune disease in this study. The study was approved by the Regional Ethical Review Board in Uppsala, Sweden. Immunohistochemistry for FoxP3
Tissue sections (4 μm thick) were made of formalin-fixed, paraffin-embedded diagnostic biopsies of PTLD. Deparaffination and antigen retrieval (Tris-ethylenediaminetetraacetic acid buffer, pH 9) were performed with a PT-link (Dako, Glostrup, Denmark). A Dako autostainer plus was used for immunohistochemical staining, including the mouse anti-human FoxP3 antibody (dilution 1:100, clone 236A/E7; eBioscience, San Diego, CA) and the EnVision DAB kit (Dako). Ten cases were stained with another mouse anti-human FoxP3 antibody (clone mAb22510; Abcam, Cambridge, MA) for comparison. Formalin-fixed, paraffin-embedded tissue sections of human tonsils were used as positive controls and stained in the same way as the PTLD tissue. The tonsil tissues were obtained from otherwise healthy patients undergoing routine tonsillectomy. Slides already stained for CD3 (n = 45) were retrieved from local pathology laboratories to estimate the frequency of CD3+ cells in each PTLD specimen. Cell Counting and Definition of FoxP3+ PTLD
Stained PTLD biopsies were assessed for FoxP3+ cells by manual counting using a light microscope (Leica DMD 108, Leica Microsystems Inc.) with the support of a grid. To create a cell-to-area ratio (expressed as of FoxP3+ cells per mm2), the area of each biopsy was calculated using the preinstalled software of the microscope. All biopsies were analyzed in their entirety, except for necrotic areas that
1041
were excluded. The analysed area of the biopsies with detectable FoxP3+ cells ranged from 1.7 to 176 mm2 (median, 44 mm2). FoxP3+ PTLD was defined as a case with 29 or more positive cells per mm2, which was the median density among the cases in which FoxP3 could be detected (Fig. 1C). The frequency of CD3+ cells was categorized into four groups based on their proportion of lymphocytes in the biopsy: 0% to less than 25%, 25% to less than 50%, 50% to less than 75%, and 75% to 100%. Statistics
The chi-square test or Fisher exact test (if
