Tumor-Infiltrating Macrophages in Post-Transplant, Relapsed Classical Hodgkin Lymphoma Are Donor-Derived

Tumor-associated inflammatory cells in classical Hodgkin lymphoma (CHL) typically outnumber the neoplastic Hodgkin/Reed-Sternberg (H/RS) cells. The composition of the inflammatory infiltrate, particularly the fraction of macrophages, has been associated with clinical behavior. Emerging work from animal models demonstrates that most tissue macrophages are maintained by a process of self-renewal under physiologic circumstances and certain inflammatory states, but the contribution from circulating monocytes may be increased in some disease states. This raises the question of the source of macrophages involved in human disease, particularly that of CHL. Patients with relapsed CHL following allogeneic bone marrow transplant (BMT) provide a unique opportunity to begin to address this issue. We identified 4 such patients in our archives. Through molecular chimerism and/or XY FISH studies, we demonstrated the DNA content in the post-BMT recurrent CHL was predominantly donor-derived, while the H/RS cells were derived from the patient. Where possible to evaluate, the cellular composition of the inflammatory infiltrate, including the percentage of macrophages, was similar to that of the original tumor. Our findings suggest that the H/RS cells themselves define the inflammatory environment. In addition, our results demonstrate that tumor-associated macrophages in CHL are predominantly derived from circulating monocytes rather than resident tissue macrophages. Given the association between tumor microenvironment and disease progression, a better understanding of macrophage recruitment to CHL may open new strategies for therapeutic intervention.


Introduction
Lymph nodes involved by classical Hodgkin lymphoma (CHL) typically contain relatively few neoplastic Hodgkin/Reed-Sternberg(H/RS) cells, which are scattered throughout a mixed inflammatory background composed of lymphocytes, macrophages, neutrophils, eosinophils and plasma cells. The H/RS cells can comprise as little as 1% of the cellularity [1]. While only the H/RS cells are neoplastic, the composition of the inflammatory background correlates with prognosis and response to treatment [2][3][4][5][6][7][8][9][10]. It is unclear to what extent the H/RS cells themselves dictate the inflammatory environment.
Of particular interest are the tumor-associated macrophages and their source as increased numbers of tumor-associated macrophages correlate with a poor prognosis in patients with CHL [11][12][13][14][15][16]. While tissue macrophages had long been presumed to be derived from circulating monocytes, recent evidence suggests that monocytes do not substantially contribute to macrophage populations in most tissues during normal homeostasis or certain inflammatory states [17]. Instead, it appears that macrophage populations within tissues maintain themselves predominantly by self-renewal, after being seeded in those locations by yolk sac precursors during embryonic development [18][19][20]. By contrast, tumor associated macrophages in the context of solid tumors [21][22][23] and inflammatory reactions associated with pathologic processes, such as atherosclerosis [24], are predominantly derived from circulating monocytes.
In addition, while self-renewal of resident macrophages may enable recovery in response to various forms of tissue injury (18), lethal radiation prior to bone marrow transplant (BMT) results in gradual replacement of tissue macrophages (with the exception of microglia) by donor monocytes in both mouse models [18,25] and human patients [26,27]. However, host macrophages may persist, particularly when fractionated doses of radiation are given, and be capable of repopulating the tissue [28]. Even when largely ablated, a small population of residual host macrophages may serve as antigen presenting cells and development of graft versus host disease [29].
The complex interplay between residual/recovering tissue macrophages and colonizing donor monocytes following BMT is of particular interest in recurrent CHL. Recurrent CHL often involves tissues with a large resident macrophage population such as lung and liver. Given the association with macrophage frequency and disease prognosis, we sought to further understand this process, including whether the inflammatory environment may change posttransplant. We characterize the inflammatory infiltrate in pre-and post-BMT tumor specimens in patients who had received an allogeneic BMT and investigate the origin of the inflammatory cells (donor or recipient) in the post-BMT specimens.

Materials and Methods
Prior to commencing the study, approval was obtained from the Johns Hopkins Hospital Institutional Review Board (JHH IRB) under the IRB-approved protocol IRB NA_00083069. Per the JHH IRB, specific oral or written consent was not required for inclusion of specimens in this retrospective study, as the findings do not affect either diagnosis or treatment and both the diagnosis and treatment plans were established prior to and independent of the study findings. The pathology database was then searched to identify patients with CHL who were treated with an allogeneic BMT at our institution and subsequently developed relapsed disease. Clinical data were reviewed by one of the authors (A.S.D.) in accordance with IRB-approved protocol. Decalcified bone marrow biopsy specimens were excluded from the study.
Formalin-fixed and paraffin-embedded (FFPE) tissue was used to perform XY chromosome studies using fluorescence in situ hybridization (FISH; CEP X SpectrumOrange/Y Spectrum-Green Direct Labeled Fluorescent DNA Probe Kit, Abbott Molecular, Des Plaines, IL) in patients who received a sex mismatched BMT. Representative fields were chosen based on the corresponding H&E stained sections, and XY FISH images were captured with a fluorescence microscope. XY FISH was also performed subsequent to standard immunohistochemical staining for CD68 on the same slide.
Molecular chimerism studies (AmpF1STR Identifiler, Applied Biosystems, Foster City, CA) were performed on the post-transplant FFPE tissue in 3 of the 4 patients. This system uses PCR and capillary electrophoresis (ABI 3130xl, Applied Biosystems, Foster City, CA) to determine the length of 15 tetranucleotide short tandem repeats (STRs) in order to generate a molecular "signature" for the patient and donor. The relative percentages of donor and patient DNA in the post-BMT CHL samples were determined using the patient and donor STR profiles that were performed prior to the BMT.

Results
Four patients were identified that had relapsed CHL after an allogeneic BMT and also had post-BMT FFPE tissue involved by CHL available in our archives ( Table 1). Two of these patients also had FFPE tissue available that was involved by CHL from before the BMT (Patients A and B).
Three patients had undergone myeloablative transplants (Patients B, C and D), and one patient (Patient A) had a non-myeloablative (reduced intensity) transplant. Three of the four patients had opposite sex donors: patients (Patients A and C) were male but had a female donor and patient (D) was a female but had a male donor. Two of the patients (B and D) had bone marrow chimerism studies available near the time of relapse. In patient B, studies performed both one month before relapse and one month after relapse (but before the initiation of treatment) showed approximately 10% patient DNA and 90% donor DNA. Patient D had 100% donor DNA with no patient DNA detected both 8 months prior to relapse and 2 months after relapse.
To assess the origin of the inflammatory infiltrate in the recurrent lymphoma we performed XY FISH on post-BMT tumor from patients A, C and D. Samples from all three patients demonstrated that the inflammatory infiltrate was largely composed of donor derived inflammatory cells (Fig 1, Fig 2, S1 Fig and S3 Fig). Scattered larger cells derived from the patient were present, which were compatible with HRS cells based on morphology and evaluation of adjacent sections stained for CD30 (Fig 1 and S1 Fig).
To confirm the FISH findings as well as to investigate the composition of the tumor in the patient with a same-sex donor (Patient B), molecular chimerism studies were performed on 3 of the 4 patients (excluding patient D). For all three, the molecular identity testing showed that the majority of the cells in the tumor were of donor origin (Patient A 89% donor; Patient B 72% donor; Patient C 62% donor) (Fig 3).
For the two patients where pre-and post-BMT tissue was available (Patients A and B), IHC stains were performed to visualize the neoplastic cells and the composition of the inflammatory infiltrate (Fig 4).  Table 2). In both patients B-cells were more numerous in the post-BMT sample, although the significance of this finding is unclear. Notably, the initial treatment for patient A included rituximab prior to BMT. The inflammatory infiltrate was also characterized for patient A at the time of the 4 th relapse, though the patient had been treated with the anti-PD-1 monoclonal antibody nivolumab prior to the biopsy (S4 Fig).

Discussion
In this study we utilized post-BMT tumor specimens in order to more fully characterize the relationship between H/RS cells and the tumor-associated inflammatory cells, in particular the origin of the tumor-associated macrophages. These findings support the idea that H/RS cells, perhaps through cytokines and chemokines, actively recruit and define the inflammatory microenvironment [30][31][32][33][34][35]. In particular, the H/RS cells appear to recruit donor-derived circulating monocytes from the bone marrow as opposed to stimulating the proliferation of resident macrophages within the tissue. For example, in Patient A, FISH showed only rare cells with patient DNA while tumor-infiltrating macrophages comprised about 15% of the cellularity in  the node (Fig 4 and S3 Fig). Macrophages that were donor-derived could also be readily identified by co-staining the same slide for CD68 by immunohistochemistry and XY FISH (Patient D, Fig 2) and appeared to comprise the vast majority of the macrophages within the lesion. Additionally, we found that the composition of the inflammatory infiltrate was very similar in the pre-and post-BMT samples where evaluable. The percentage of macrophages for the two patients examined in this manner fell into the "intermediate risk" category as defined by Steidl, et al. with between 5 and 25% macrophages [13] and similar percentages in the relapsed   Fig 1A (white box) is shown at higher magnification. Prior to analysis by XY FISH, the same slide was stained for CD68 using standard immunohistochemical techniques in order to identify macrophages (A). The images of the DAPI nuclear stain and the CD68 cytoplasmic stain are overlaid (B) to better identify individual cells in the corresponding FISH images (C). Where possible to discern, the tumor-infiltrating macrophages are all derived from the male donor (arrows, XY, red and green). However, by this method we found that in 21 +/-4% of DAPI-positive nuclei, it was not possible to score X, Y status due to sectioning and/or other technical limitations. Of note, this patient had 100% donor chimerism in her bone marrow when tested near the time of this biopsy. H/RS cells (Fig 1 and S1 Fig)   specimens post-BMT. The percentage of CD8-positve T-cells also remained similar pre-and post-BMT, suggesting that the H/RS cells are recruiting a specific complement of inflammatory cells. Of note, EBV may affect the inflammatory environment by altering cytokine expression [36] and has been associated with increased CD8+ T-cells, increased LAG-3-positive regulatory T-cells and increased macrophage recruitment [11,37]. However, in this series, only one of the four cases was EBV-associated, and a corresponding primary diagnostic specimen was not available for comparison in this patient. Even in the absence of EBV, studies of the inflammatory environment in CHL have demonstrated an association with immunosuppressive molecules and an impaired T-cell response [37,38]. For example, increased expression of immune modulators such as indoleamine 2,3-dioxygenase (IDO) was associated with worse clinical outcome. IDO expression strongly correlated with extent of macrophage infiltration and was expressed by macrophage/monocyte-derived cells [38]. While it is possible that BMT may result in an alteration in the tumor microenvironment to a more pro-inflammatory state with graft versus tumor effect, the similar inflammatory environments pre and post-transplant in terms of percent of tumor-infiltrating macrophages, and fractions of CD4 and CD8+ T-cells argue against this possibility. Nonetheless, this does not rule out the possibility that some patients may have a favorable change in their immune response post-transplant.
One potential limitation on the generalizability of this study for understanding tumor-infiltrating macrophage recruitment to the CHL microenvironment is the demonstration in prior studies that total body irradiation is associated with gradual replacement of tissue macrophages by bone marrow-derived monocytes [18,[25][26][27]. However, our study did include one patient who had undergone a non-myeloablative BMT (Patient A), and even in that patient, the inflammatory infiltrate remained predominantly donor-derived (S3 Fig). This suggests the recruitment of bone marrow derived monocytes to recurrent CHL is not a mere artifact of the BMT conditioning regimen.
While the number of cases is small, our findings raise the possibility of novel treatment strategies for recurrent CHL, including the potential to pharmacologically inhibit monocyte recruitment to tumor lesions to slow or alter disease progression [39,40].   1). XY FISH revealed that large atypical nuclei were predominantly derived from the female patient (XX, red). In addition, there were smaller patient-derived nuclei (XX, red) with a distribution consistent with endothelial cells. However, the majority of the surrounding inflammatory cells appeared donor-derived (XY, red and green). This corresponded to 78% of total nuclei (as highlighted by DAPI), including only cells Nuclei are stained with DAPI (blue). Red and green signals correspond to the probes on the X and Y chromosomes, respectively. The pre-BMT tissue (cervical lymph node) demonstrates both X and Y chromosomes in all cells whereas the post-BMT tissue (portal lymph node) shows that most cells are XX (red; donor, 85% of those that could be identified) with only scattered large XY (red and green; patient, 15% of those that could be identified) cells. In 21% of nuclei, the X, Y FISH probes could not be definitively evaluated due to the plane of section.