The Clinical Significance and Prognostic Role of Whole-Blood Epstein-Barr Virus DNA in Lymphoma-Associated Hemophagocytic Lymphohistiocytosis

To evaluate the role of circulating Epstein-Barr virus (EBV) DNA in lymphoma-associated hemophagocytic lymphohistiocytosis (HLH). We conducted a retrospective cohort study to explore the clinical and prognostic significance of EBV DNA in lymphoma-associated HLH. We included adult patients with combined diagnoses of lymphoma and HLH from January 2010 and November 2022 by retrieving the medical record system. A total of 281 patients with lymphoma-associated HLH were identified. Elevated whole-blood EBV DNA was observed in 54.4% (153/281) of patients, and the median copy number was significantly higher in the T/NK-cell malignancies (199,500, interquartile range, 30,000–1,390,000) than that in the B-cell non-Hodgkin lymphoma (5520, interquartile range, 1240–28,400, P < 0.001). The optimum cutoff for predicting survival was 16,100 copies/mL. Compared to the patients with EBV DNA ≤ 16,100 copies/mL, those with EBV DNA > 16,100 copies/mL were younger and had more T/NK-cell malignancies, lower levels of neutrophils and fibrinogen, and higher levels of hemoglobin, alanine aminotransferase, aspartate aminotransferase, lactic dehydrogenase, and β2-microglobulin. A higher load of EBV DNA (> 16,100 copies/mL), thrombocytopenia (< 100 × 109/L), neutropenia (< 1 × 109/L), hypofibrinogenemia (≤ 1.5 g/L), and elevated levels of creatinine (> 133 μmol/L) were independent adverse predictors of 60-day overall survival and overall survival. A prognostic index based on EBV DNA and the other four factors was established to categorize the patients into four groups with significantly different outcomes. Our study identified high EBV load as a risk factor for lymphoma-associated HLH and established a prognostic index to predict outcomes.


Introduction
Hemophagocytic lymphohistiocytosis (HLH) is a potentially devastating disorder that is characterized by overwhelming inflammation [1]. HLH can be classified as familial HLH and non-familial HLH according to the underlying etiologies [2]. The common secondary causes include malignancies, infections, and autoimmune disorders. For malignancies underlying HLH, lymphoma is the most common pathological type. The common subtypes of lymphoid malignancies include diffuse large B-cell lymphoma (DLBCL), aggressive NK-cell leukemia (ANKL), and extranodal NK/T-cell lymphoma, nasal type (ENKL) [3]. The reported outcomes for patients Jing Zhang, Shuchao Qin, and Ze Jin have contributed equally to this work. with lymphoma-associated HLH varied across different studies, with median survival ranging from 2 to 6 months [4][5][6][7][8]. Although generally considered as a disease with a dismal prognosis, lymphoma-associated HLH shows heterogeneous outcomes with some patients having long-term disease-free survival while others succumbing to the disease rapidly despite aggressive therapy. Risk stratification of lymphoma-associated HLH may guide the treatment selection and could potentially improve the outcomes of patients with lymphoma-associated HLH. However, until now, the studies focusing on the prognostication of lymphoma-associated HLH were very limited and there were no recognized prognostic tools for lymphoma-associated HLH.
Epstein-Barr virus (EBV) is a common herpesvirus that mostly remains latent following acute infection and could be involved in the development of both familial and secondary HLH [9]. EBV-associated non-familial HLH could be purely caused by EBV infection or associated with an underlying malignancy. Malignancies underlying EBV-associated HLH include T/NK-cell chronic active EBV infection and EBV-positive lymphoma. In the latter scenario, ANKL, ENKL, and EBV-positive DLBCL are the most common subtypes [3].
The quantification and dynamic evaluation of circulating EBV DNA play an important role in patients with EBV-associated malignancies [10]. Elevated pretreatment whole-blood EBV DNA has been demonstrated as a negative prognostic factor in patients with Hodgkin lymphoma, ENKL, angioimmunoblastic T-cell lymphoma (AITL), or chronic lymphocytic leukemia [10][11][12][13]. Elevated whole-blood EBV DNA could be frequently observed in cases of lymphoma with HLH; however, the clinical significance and prognostic role of whole-blood EBV DNA have not been studied in lymphoma-associated HLH. In the current study, by analyzing a large cohort of lymphoma-associated HLH, we investigated the clinical significance and prognostic role of circulating EBV DNA. Furthermore, by combining circulating EBV DNA and other prognostic factors, we established a prognostic index that could categorize patients with lymphoma-associated HLH into four groups with significantly different outcomes.

Study Population
By retrieving the medical record system, adult patients with lymphoma-associated HLH diagnosed at the First Affiliated Hospital of Nanjing Medical University from January 2010 to November 2022 were included in this study. The inclusion criteria were as follows: (1) age ≥ 18 years; (2) fulfillment of five or more of the eight HLH-2004 diagnostic criteria; and (3) histopathological diagnosis of lymphoma. Exclusion criteria include (1) primary HLH caused by germline genetic abnormalities and (2) patients without baseline EBV data. The volume of newly diagnosed lymphomas per year in our hospital is shown in Supplementary Fig. S1. In most circumstances, HLH was suspected when patients presented with persistent fever and/or cytopenia. We also suspected HLH when hemophagocytosis was present in the bone marrow examination for the staging of lymphoma. Other diagnostic tests including ferritin and soluble CD25 were further evaluated to establish the diagnosis of HLH.

Determination of EBV DNA
The peripheral whole blood samples of patients were collected in an EDTA-containing tube. DNA was extracted with the EBV PCR Fluorescence Quantitative Diagnostic Kit (Daan Gene Co., Guangzhou, China). Quantification of EBVspecific sequences was performed by real-time quantitative PCR with an ABI PRISM 7500 (Applied Biosystems, Foster City, CA, USA). The cutoff was 500 copies/mL.

Data Collection
For prognostic analysis, the latest laboratory values measured before a specific HLH-directed or lymphoma-directed therapy were used. When no data was available, the first value within 72 h after treatment initiation was notated. Follow-ups were conducted by their clinical records or by telephone calls to the patients or their family members.

Statistical Analyses
The D'Agostino-Pearson test was used for examining the normality of continuous variables. If the values passed the normality test and equal variance test, we used Student's t-test; otherwise, we used the Mann-Whitney test. Fisher's exact test or chi-square test was used to compare categorical variables. Overall survival (OS) was calculated from the time of diagnosis to death or the last follow-up. The best cutoff of EBV DNA was determined by X-tile, version 3.6.1 (Yale University). We generated survival curves via the Kaplan-Meier method and log-rank test and analyzed predictors of survival using the univariate and multivariate Cox regression analyses. Statistical analyses were performed using STATA/MP, version 17.0 (StataCorp LLC) and GraphPad Prism, version 8.2.1 (GraphPad Software). P < 0.05 (2-sided) was considered statistically significant.

Clinical Characteristics
A total of 281 cases of lymphoma-associated HLH were identified, and the incidence of HLH among patients with lymphoma per year ranged from 0.01 to 0.06 (Supplementary Fig. S2). The median age at diagnosis was 54 years old. The male-to-female ratio was 2.1:1. T/NKcell malignancies (57.7%, 162/281) were the most common subtypes, followed by B-cell non-Hodgkin lymphoma (B-NHL, 39.9%, 112/281) and Hodgkin lymphoma (2.1%, 6/281). One patient had peripheral T-cell lymphoma concomitant with chronic lymphocytic leukemia. In the cohort, 236 (84.0%) patients had HLH and lymphoma diagnosed simultaneously, while 44 (15.7%) patients had HLH with relapsed or refractory lymphoma and one patient had HLH during the treatment of lymphoma. EBV (54.4%, 153/281) and hepatitis B virus (defined as hepatitis B surface antigen positive or hepatitis B core antibody positive, 38.0%, 104/274) infections were common in adult lymphoma-associated HLH. Fifty (17.8%) patients received only HLH-directed treatments like HLH-1994/2004 regimen, 98 (34.9%) patients were directly exposed to lymphoma-directed treatments like CHOP ± rituximab or EPOCH ± rituximab, 110 (39.1%) patients received both, and 23 (8.2%) patients received neither treatment (Table 1). At a median follow-up of 383 days, 174 (61.9%) patients died. The median OS of the whole population was 103 days. The 60-day OS rate was 55.4%.

Whole-Blood EBV DNA Levels and the Cutoff for OS
Among the 153 patients with elevated whole-blood EBV DNA copy number, the median copy number was 98,200 (interquartile range, 12,100-999,000) copies/mL, significantly higher in the T/NK malignancies (199,500, interquartile range, 30,000-1,390,000) than that in the B-NHL cases (5520, interquartile range, 1240-28,400, P < 0.001). As patients with elevated EBV DNA copy number had worse 60-day OS rate (50.0% vs 61.7%, log-rank P = 0.0347) and OS (median 61 days versus 236 days, log-rank P = 0.0125) compared with those with normal EBV DNA level, we determined the optimum cutoff of EBV DNA copy number for 60-day OS, which was 16,100 copies/mL, by the X-tile software for further analysis (EBV DNA < 500 copies/mL was calculated as 500 copies/mL). The detailed results are shown in Supplementary Fig. S3.

Differences in Clinical and Laboratory Features Between Groups Based on EBV DNA
To evaluate the characteristics of lymphoma-associated HLH with different EBV DNA loads, we compared variables of interest between the groups divided by the optimum cutoff of EBV DNA copy number. In the cohort, 171 (60.9%) patients had EBV DNA ≤ 16,100 copies/mL and 110 (39.1%) had EBV DNA > 16,100 copies/mL. The patients with EBV DNA > 16,100 copies/mL were younger and had more T/NK-cell malignancies than those with EBV DNA ≤ 16,100 copies/mL. Additionally, lower levels of neutrophils and fibrinogen along with higher levels of hemoglobin, alanine aminotransferase, aspartate aminotransferase, lactic dehydrogenase, and β 2 -microglobulin were seen in patients with a higher EBV DNA load ( Table 2).

Development of a Prognostic Index of Survival
We developed a prognostic index of 60-day OS and OS based on the risk factors identified by the multivariate analysis, with one point given for fulfilling each of the following criteria: EBV DNA > 16,100 copies/mL, platelets count < 100 × 10 9 /L, neutrophils count < 1 × 10 9 / L, fibrinogen ≤ 1.5 g/L, and creatinine > 133 μmol/L. The total score was obtained by adding, ranging from 0 to 5. Among the 273 patients with complete data of the five variables, 18 patients scored 0, 101 scored 1, 76 scored 2, 57 scored 3, 20 scored 4, and 1 scored 5.
There were significant differences in 60-day OS and OS between the patients who scored 1 and 2 (log-rank P < 0.001 and P = 0.0016), 2 and 3 (log-rank P = 0.0164 and P = 0.0050), as well as 3 and 4 (log-rank P = 0.0096 and P = 0.0153). No significant difference in 60-day OS and OS was observed between the patients who scored 0 and 1 (log-rank P = 0.3322 and P = 0.3838) as well as 4 and 5 (log-rank P = 0.0665 and P = 0.0665). Hence, we divided the patients into four groups: low-risk (score ≤ 1), intermediate-risk (score = 2), high-risk (score = 3), and very high-risk (score ≥ 4). The 60-day OS rate was 78.4% in the low-risk group, 50.3% in the intermediaterisk group, 33.2% in the high-risk group, and 4.8% in the very high-risk group, with a significant trend (P < 0.001; Fig. 2A). The median OS time was 405 days in the lowrisk group, 91 days in the intermediate-risk group, 21 days in the high-risk group, and 13 days in the very high-risk group, with a significant trend (P < 0.001; Fig. 2B).

Discussion
In the current study, we analyzed the clinical and prognostic significance of EBV DNA in lymphoma with HLH. With the X-tile software, 16,100 copies/mL was determined as the best cutoff for EBV DNA in predicting survival in lymphoma with HLH. And we found that elevated whole-blood EBV DNA (> 16,100 copies/mL) significantly predicted reduced short-and long-term survival and was remarkably associated with high-risk clinical and laboratory features. Further multivariate analysis demonstrated that elevated whole-blood EBV DNA was an independent predictor in the prognostication of lymphoma with HLH. Finally, a prognostic index based on EBV DNA and four other factors was established and demonstrated robustness in predicting the outcomes of patients with lymphoma-associated HLH. EBV is involved in the pathogenesis of both lymphoma and HLH. For lymphoid malignancies including NK-cell lymphomas and EBV-positive DLBCL, EBV infects the tumor cells and plays an important role in the development of these lymphoid malignancies [14,15]. For AITL, EBV is detected in non-tumor B cells; however, these B cells are crucial for the pathogenesis of AITL. EBV may contribute to the pathogenesis of AITL by modulating and transforming these non-tumor B cells [16]. For EBV-positive lymphomas, a higher EBV load may trigger or aggravate HLH. For instance, in patients with EBV-positive NK-cell lymphoid tumors including ANKL and ENKL, a higher EBV load was significantly associated with the presence of HLH [17]. However, it should be noted that factors other than EBV are also very important in the pathogenesis of HLH in these NK-cell lymphoid malignancies, as HLH is also frequently present in patients with EBV-negative ANKL [18]. EBV may contribute to the disease severity of lymphomaassociated HLH from several aspects. Firstly, elevated EBV DNA is associated with elevated IFN-γ, which predicts a poorer prognosis in patients with HLH [19]. Therefore, EBV may lead to a more severe hyperinflammatory status, thereby contributing to a poorer prognosis for patients with lymphoma-associated HLH. Secondly, as in nasopharyngeal carcinoma and EBV-associated lymphoma, a high EBV DNA copy number is associated with a high tumor burden, which usually predicts a worse prognosis [20]. In our cohort, elevated EBV DNA was associated with a higher lactic dehydrogenase level, suggesting that the EBV DNA copy number correlates with the tumor burden. Thirdly, a higher EBV DNA copy number was more frequently observed in patients with T/NK cell lymphoma, which usually have an inferior prognosis. Other mechanisms remain to be explored in future studies.
The prognostic factors for lymphoma-associated HLH have been studied in several previous studies. Zhao et al. found that the NK-cell type lymphoid malignancy was a significant predictor in patients with lymphoma-associated HLH [20]. In the study by Yao et al., which included 104 non-Hodgkin lymphoma-associated HLH, age, international prognostic index, platelet count, and the NK/T lymphoma subtype were independent risk factors for non-Hodgkin lymphoma-associated HLH [21]. Other factors, including poor performance status and hyponatremia, were identified as negative prognostic factors for lymphoma-associated HLH [4,22]. Li et al. also found that a higher EBV load was a negative predictor in B-NHL-associated HLH; however, this study was restricted to B-NHL and only included 31 cases [4]. In our study, a higher EBV load was a risk factor for T/NK-cell lymphoma-associated HLH but not B-NHLassociated HLH. Although T/NK-cell lymphoma was a significant predictor of worse prognosis in lymphomaassociated HLH, after multivariate analysis, it was not an independent risk factor. These results suggest the EBV load is more important than the lymphoma subtype in determining the prognosis of lymphoma-associated HLH. The prognostic index based on EBV DNA contains common laboratory parameters, which are simple and feasible in clinical application. External validation is necessary for the reliability of the predictive models. We need to validate this prognostic index in a multi-center study in the near future. The poor prognosis of lymphoma-associated HLH with a higher EBV DNA copy number or a higher prognostic index suggests these cases should be treated with more effective regimens. As in our study, traditional regimens including the HLH-1994/2004 regimens, CHOP, and EPOCH could not improve the prognosis in patients with a higher EBV load. The strategies designed to control EBV-associated HLH could be used to treat lymphoma-associated HLH with a higher EBV DNA copy number. Rituximab could target EBV-infected B cells and reduce the EBV load, thereby controlling EBV-associated HLH. Therefore, rituximab could also be effective in treating lymphoma with HLH and a higher EBV load. However, rituximab is not effective for cases in which EBV infects T cells and/or NK cells [23]. The L-DEP regimen is effective in EBV-associated HLH in pediatric and adult patients [24,25]. Therefore, the L-DEP regimen could be useful in managing lymphoma-associated HLH with a higher EBV load. Additionally, the PD-1 antibody nivolumab showed promising efficacy in relapsed/ refractory EBV-associated HLH. According to the study by Liu et al., in 7 patients with relapsed/refractory EBV-associated HLH who received nivolumab treatment, five patients achieved complete remission [26]. Hence, the anti-PD-1 therapy could be used to manage lymphoma-associated HLH with a higher EBV load, especially in patients with ENKL, as the anti-PD-1 therapy is highly effective in patients with ENKL [27]. Future clinical trials may evaluate these regimens in lymphoma-associated HLH with a higher EBV load. Our study has several limitations. This is a single-center retrospective study, although we have a relatively large cohort. The treatments used in our cohort are not uniform and could be a confounding factor in determining other prognostic factors for lymphoma-associated HLH. The study by Zheng et al. has demonstrated that plasma EBV DNA was superior to EBV DNA from peripheral blood mononuclear cells in monitoring EBV-associated HLH [17]. However, in our study, only the whole-blood EBV DNA was determined and whether plasma EBV DNA is a better prognostic factor remains to be explored.

Conclusions
The current study identified high EBV load as a risk factor for lymphoma-associated HLH. A prognostic index based on EBV DNA and the other four factors could predict patients' outcomes and potentially guide the treatment selection.
Author Contribution Jing Zhang, Shuchao Qin, and Ze Jin: conceptualization, methodology, data curation, formal analysis, writing-original draft, and visualization. Qingqing Chen, Lingxiao Xing, and Tonglu Qiu: data curation, formal analysis, and visualization. Yi Xia and Jinhua Liang: data curation, formal analysis, and supervision. Huayuan Zhu, Li Wang, and Lei Fan: conceptualization, writing-review and editing, and supervision. Wei Xu, Jianyong Li, and Yi Miao: conceptualization, methodology, funding acquisition, supervision, writing-review and editing. All authors approved the final version of the manuscript for publication.

Data Availability
The data that support the findings of this study are available on request from the corresponding author.

Declarations
Ethics Approval The study was approved by the Ethics Committee of Jiangsu Province Hospital.

Consent to Participate
Informed consent was obtained from all individual participants included in the study.

Consent for Publication Not applicable.
Competing Interests The authors declare no competing interests.