Pretreatment EBV-DNA Copy Number Is Predictive of Response and Toxicities to SMILE Chemotherapy for Extranodal NK/T-cell Lymphoma, Nasal Type

Epstein–Barr virus (EBV) causes a variety of benign and neoplastic diseases, including infectious mononucleosis, posttransplantation lymphoproliferative disorder (PTLD) and EBV-associated malignancies such as lymphomas, including extranodal NK/T-cell lymphoma, nasal type (ENKL), Hodgkin lymphoma, Burkitt lymphoma, age-associated large B-cell lymphoma, and several other T-cell lymphomas (1–3). ENKL is a rare subtype of non-Hodgkin lymphoma, mainly occurs in the nasal/paranasal area, skin, or gastrointestinal tract and is much more common in Asia and Latin America than in Western countries (4–6). The prognosis of ENKL was poor under conventional radiotherapy and/or chemotherapy (4, 7) but has recently improved by concurrent chemoradiotherapy (8, 9) or newly developed SMILE chemotherapy, comprising the steroid dexamethasone, methotrexate, ifosfamide, l-asparaginase, and etoposide (10, 11).

This type of lymphoma is invariably associated with episomal infection of EBV in the tumor cells, which implies its tumorigenic role. The presence of EBV-DNA in peripheral blood has been used as a surrogate marker for estimating tumor amount in several EBV-associated malignancies (12–14). In particular, after organ transplantation, increasing loads of EBV in whole blood, lymphocytes, and plasma are associated with corresponding increases in the risk of PTLD (14). For nasopharyngeal carcinoma, plasma EBV-DNA load is known to be useful for monitoring disease activity and predicting the outcome of treatment (15, 16). The disease activity and prognosis of ENKL can also be monitored by measuring circulating EBV-DNA in plasma (17). For patients registered to the SMILE phase II study, we simultaneously conducted a prospective observational study, SMILE-EBV study, in which the amounts of EBV-DNA in whole blood and plasma were evaluated for ENKL.

The aim of this study is to evaluate the copy numbers of EBV-DNA from whole blood and plasma in patients with ENKL who received SMILE chemotherapy. The predictive value of EBV-DNA for tumor response, toxicity, and prognosis was analyzed, as well as the preference of samples from whole blood or plasma. The eligibility criteria, treatment, and response were described in the report of the phase II study (11). EBV-encoded small RNA (EBER) in situ hybridization positivity was counted in accordance with our previous study (18). A total of 38 patients were enrolled: 26 from Japan, 6 from Hong Kong, and 6 from South Korea. The amounts of EBV-DNA were measured in whole-blood and plasma samples from patients participating in this phase II study at 3 time points: before the treatment, after 2 courses of SMILE and after a series of treatments. Because of the lack of an international standardized method for quantification of EBV-DNA, the 26 patients from Japan were the subjects for this study. All samples were measured in a central laboratory. Registration onto the study was conducted by facsimile from the participating institutes to the C-SHOT Data Center (Nagoya, Japan), simultaneously with the entry into the SMILE study. The study was approved by both the Protocol Review Committee and the Institutional Review Board of each institution in Japan. Written informed consent was obtained from all of the patients. The study was registered with the University Hospital Medical Information Network Clinical Trials Registry (UMIN-CTR number, UMIN000000741), as an associated but separate study of SMILE phase II (UMIN-CTR number, UMIN000000712).

Complete response was defined as the complete disappearance of all objective signs of disease, including enlarged lymph nodes or hepatomegaly and splenomegaly at restaging. Partial response was defined as at least a 50% reduction of tumor volume without the occurrence of new lesions at restaging. Progressive disease was defined as a greater than 25% increase in the sum of tumor lesions or the emergence of one or more new lesions or clinical symptoms that indicate disease progression. No response was defined as any response that did not fall into the categories defined above. If a patient died of any cause before day 42 of the second course of SMILE and could not undergo the defined restaging procedure, the patient's response was recorded as early death. The overall response rate (ORR) was defined as the proportion of all patients who could be evaluated for response who experienced complete or partial response. Toxicity was graded according to the Common Terminology Criteria for Adverse Events v3.0.

A 5-mL patient peripheral blood was obtained, sent to the central laboratory (Nagoya University Graduate School of Medicine), and divided into whole-blood and plasma samples. DNA was extracted from 200 μL of either whole blood or plasma, using QIAamp DNA blood kits (Qiagen K.K.). A real-time quantitative PCR assay was carried out and the result was expressed as copies per 1 mL of sample, as previously described (19, 20). The minimum detection level was 2 copies per reaction that was equivalent to 100 copies/mL for whole blood or plasma.

Regression analysis compared the copy numbers in whole blood and plasma. Fisher exact test was used to compare the responses or toxicities to the SMILE chemotherapy. Mann–Whitney U test and Kruskal–Wallis test were used to compare the levels of EBV-DNA between patient groups. Cut-off value of the categorization by EBV-DNA levels were determined by the receiver operating characteristic analysis. Patient survival data were analyzed by the method of Kaplan and Meier and were compared by log-rank test. Univariate and multivariate analyses were carried out using Cox proportional hazard model. Data were analyzed with STATA version 11 and SPSS (SPSS) software.

The baseline characteristics of 26 eligible patients are listed in Table 1. The median age was 46.5 (range: 17–67) years, and the male:female ratio was 14:12. Twelve patients (46%) had newly diagnosed stage IV disease, 11 were in first relapse, and 3 were in the primary refractory status. EBER in situ hybridization was positive in all specimens, with a median positivity of 68% (range: 12%–96%) of lymphoma cells.

EBV-DNA was detected in 22 samples of whole blood (median: 3.7 × 10³, range: 0–1.1 × 10⁷ copies/mL) and 15 samples of plasma (median: 8.7 × 10², range: 0–1.3 × 10⁷ copies/mL). The level of EBV-DNA was not different among the 3 disease state (newly diagnosed, relapsed or refractory) groups at enrollment both in whole blood (P = 0.19 by Kruskal–Wallis test) and in whole blood (P = 0.22). An inconsistent result was seen in 9 patients. EBV-DNA was positive in whole blood but was negative in plasma in 8 patients. Conversely, in another patient, the EBV-DNA was only detected in plasma. EBV-DNA was not detected in either whole blood or plasma in 3 patients (nos. 9, 23, and 25). The concordance rate between whole blood and plasma was 65% (17 of 26). The viral DNA copy numbers were compared between whole blood and plasma before SMILE chemotherapy. A strong correlation was found between the amounts in whole blood and those in plasma (r = 0.997, P < 0.001, Fig. 1). No differences were found for the EBV-DNA level among patients with newly diagnosed stage IV, relapsed and refractory status (P = 0.19 for whole blood and P = 0.24 for plasma). No significant correlation was found between EBER positivity and plasma or whole blood EBV-DNA level (Supplementary Fig. S1).

EBV loads in whole blood or plasma from the 16 patients (8 with complete response, 7 with partial response, and 1 with progressive disease) were measured before the treatment, after 2 courses of SMILE chemotherapy, and after a series of treatments. Viral load declined in most patients with complete response or partial response after 2 courses of SMILE chemotherapy and/or after a series of treatments (Fig. 2). However, 5 patients with complete or partial response did not show the decrease of viral load. Of these, 2 patients experienced disease recurrence, and another patient died of transplant-related mortality in complete response. Other 2 patients maintained response at the time of last follow-up.

Among the 26 patients, there were 12 patients with complete response, 8 with partial response, 1 with no response, 3 with progressive disease, and 2 with early death (Table 2), and the ORR was 77%. For patients with less than 10⁵ copies/mL of EBV-DNA in whole blood, the ORR was 90% (19 of 21), but was 20% (1 of 5) in patients with 10⁵ copies/mL or more (P = 0.005). In addition, the ORR was 95% (18 of 19) in patients with less than 10⁴ copies/mL of EBV-DNA in plasma, but was 29% (2 of 7) in patients with 10⁵ copies/mL or more (P = 0.002). All 3 patients without detectable EBV-DNA in either whole blood or plasma attained complete response. The amounts of EBV-DNA before treatment were not significantly different between patients with complete response and those with partial response (whole blood, P = 0.82; plasma, P = 0.68).

Grade 4 leukopenia (77%) and neutropenia (88%) were commonly observed. Grade 4 anemia was encountered in one patient and thrombocytopenia was seen in 9 patients. The nonhematologic grade 4 toxicities included infection (n = 2), alanine aminotransferase elevation (n = 1), and encephalopathy (n = 1); 3 patients experienced grade 4 somnolence, which was complicated by a grade 3 infection in one patient and by grade 4 encephalopathy in another patient. One patient experienced grade 2 pancreatitis and had complications from grade 4 hyponatremia, hyperamylasemia, and appetite loss. Grade 4 toxicity other than leukopenia/neutropenia was significantly higher in patients with 10⁵ copies/mL of EBV-DNA or more in whole blood (100% vs. 29%, P = 0.007). Grade 4 toxicity other than leukopenia/neutropenia was also significantly higher in patients with 10⁴ copies/mL of EBV-DNA or more in plasma (86% vs. 26%, P = 0.002; Table 2).

Patients with 10⁵ copies/mL of EBV-DNA or more in whole blood showed significantly lower survival than those with less than 10⁵ copies/mL (Fig. 3A, P < 0.0001). Similarly, the prognosis of patients with 10⁴ copies/mL of EBV-DNA or more in plasma was significantly worse than that in those with less than 10⁴ copies/mL (Fig. 3B, P < 0.0001). EBER positivity of more than 75% was also a factor associated with poor prognosis (Fig. 3C). Plasma and whole blood EBV-DNA before SMILE chemotherapy were significant prognostic factors for overall survival by univariate analysis, as well as serum lactate dehydrogenase (LDH) elevation, B symptom, and EBER positivity (Table 3). Multivariate analysis showed that LDH elevation [HR, 8.5; 95% confidence interval (CI), 1.9–38.0] and pretreatment whole blood EBV-DNA (HR, 65.5; 95% CI, 5.3–813.7) were significant prognostic factors. Plasma EBV-DNA was not prognostic (HR, 3.90; 95% CI, 0.70–21.8) if adjusted by LDH elevation using multivariate analysis. EBER positivity showed marginal significance (HR, 3.3; 95% CI, 0.95–11.8) if included in the model with LDH elevation.

For EBV-associated malignancies, the significance placed on EBV-DNA in peripheral blood as a biomarker has increased in recent decades. Previous studies have reported that the level of EBV-DNA in a peripheral blood compartment is a useful biomarker in EBV-associated malignancies (14, 21). Lei and colleagues found a significant reduction of plasma EBV-DNA in patients with EBV-associated lymphoid malignancies (Hodgkin lymphoma, nasal NK/T cell lymphoma, PTLD, and Burkitt lymphoma) during the course of effective therapy (21). In addition, disease progression was associated with a rapid increase in plasma EBV-DNA levels in patients with ineffective therapy. Gandhi and colleagues showed that EBV-DNA is specifically detected in plasma of EBV-positive Hodgkin lymphoma patients before treatment (22). Viral DNA was undetectable following therapy in responsive patients and patients with long-term remission. Patients who experienced relapse had a significantly higher plasma EBV-DNA concentration before treatment. The plasma DNA concentration was persistently low or undetectable in patients with complete clinical remission. Overall survival and relapse-free survival were significantly higher for patients with a pretreatment plasma EBV-DNA level of less than 1,500 copies/mL. Au and colleagues reported that plasma EBV-DNA level is valuable as a tumor biomarker and for prognostication in EBV-positive lymphoma (17). EBV-DNA in plasma became undetectable for patients in remission but was elevated for those with refractory disease. A high level of EBV-DNA was significantly associated with inferior overall survival by multivariate analysis. Subgroup analysis of NK cell lymphoma showed that the level of EBV-DNA was also correlated with disease stage. Presentation of a high level of EBV-DNA was also significantly associated with inferior overall survival by multivariate analysis in their cohort. Prognostic factors of lymphoma may change when the treatment modality changes (23). In this study, however, EBV-DNA copy number in plasma or whole blood was also predictive of response and survival of ENKL patients who received SMILE chemotherapy, in agreement with other observations in the literature.

Another novel finding is that severe adverse events of the chemotherapy were also predictable using the EBV-DNA amount, which has not been identified by other studies in the literature. This analysis is only possible by examining patients who receive exactly the same treatment, ideally subjects of prospective studies. Because the level of EBV-DNA was not different by the 3 disease status groups (newly diagnosed, relapsed, or refractory), we examined the patients together in this study. As an interpretation of this finding, patients with higher tumor burden may experience more severe toxicity because of poor general condition or tissue damage by the tumor. Another hypothesis is that the toxicity by chemotherapy is mediated by certain toxic substances in tumor cells. Because NK cells possess cytotoxic activities, almost all ENKL have cytotoxic molecules such as perforin or granzymes. In several EBV-associated malignancies, the high viral load may be explained by the tumor releasing viral components (24, 25). Toxic substances that are released from tumor cells degraded by chemotherapy such as SMILE, although they may not be cytotoxic molecules, may contribute to the high rate of adverse reactions after chemotherapy. Whatever the reason, measurement of EBV-DNA may be helpful for patient stratification to avoid excessive toxicity because the myelosuppressive adverse reaction of SMILE is rather profound for a part of patients.

Plasma is used as samples in most studies for evaluating EBV-DNA as a biomarker in EBV-associated disease (13–17, 22). However, controversies exist as to which blood compartment should be used for measuring EBV because several compartments of blood, whole blood, peripheral blood mononuclear cells, plasma, and serum can be used in the studies. Our previous study compared the usefulness of plasma and mononuclear cells for detecting EBV-DNA in ENKL patients, although the treatment was not unified (16). For the diagnosis of EBV-associated PTLD, earlier studies used peripheral blood mononuclear cells because EBV infection occurs in this cell compartment (26, 27). Plasma or serum samples are readily obtained and widely used for diagnosing EBV-associated PTLD; however, the sensitivity seemed to be low (28, 29). Several reports have revealed that whole blood, containing both cellular and humoral compartments, is better than plasma/serum when testing patients with PTLD (20, 30, 31). Recently, Spacek and colleagues reported that plasma is better than whole blood for the monitoring and estimation of prognosis for Hodgkin lymphoma (32). Plasma samples may be recommended as a biomarker of disease activity rather than peripheral blood mononuclear cells in patients with Hodgkin lymphoma, as shown in another study (20). However, comparison among each blood compartment has not been well investigated. Useful compartments may differ among diseases (14). In this study, the levels of EBV-DNA in plasma were compared with those in whole blood. Although strong correlation was detected between the viral copy numbers in whole blood and those in plasma, EBV-DNA was more frequently detected in whole-blood samples before treatment. Notably, EBV-DNA was only detected in whole blood in 8 patients, whereas it was only positive in plasma in one patient. This suggests that whole blood is more suitable than plasma to examine the EBV-DNA for ENKL. The reason for the phenomenon that EBV-DNA was only detected in whole blood remains undetermined. Among such 8 patients in this study, only 4 patients showed bone marrow involvement, and none accompanied leukemic presentation. The only possible explanation is that EBV-DNA might be lost or degraded in the fractionation procedure. Another point of interest is that EBV-DNA was not detected in either whole blood or plasma in 3 patients, although EBER was positive in tissue samples. Therefore, EBV-DNA detection in peripheral blood cannot be used as an alternative to the histologic detection of EBV or the diagnosis of ENKL. Moreover, the levels of 10⁵ copies/mL of EBV-DNA in whole blood and 10⁴ copies/mL of EBV-DNA in plasma seem to be cut-off values: the patients with copy numbers lower than these showed significantly better outcome. These 2 copy numbers also showed clinical value to predict severe adverse events.

In conclusion, our study indicates that the level of EBV-DNA in plasma or whole blood can predict response and adverse events of SMILE chemotherapy for newly diagnosed stage IV, relapsed, or refractory ENKL. Whole-blood samples were more suitable for this purpose, although plasma was preferable for other purposes such as diagnosis of EBV infection.

R. Suzuki received honoraria from Kyowa-Hakko Kirin Company. K. Oshimi is currently an employee of Eisai Pharmaceutical Co., Ltd. (Tokyo, Japan). No potential conflicts of interest were disclosed by the other authors.

Conception and design: M. Yamaguchi, J. Suzumiya, K. Kawa, K. Oshimi, R. Suzuki

Development of methodology: H. Kimura, K. Oshimi

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Ito, Y. Maeda, C. Hashimoto, F. Ishida, K. Izutsu, N. Fukushima, Y. Isobe, Y. Hasegawa, S. Okamura, H. Kobayashi, R. Hyo

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Ito, F. Ishida, M. Yamaguchi, K. Oshimi, R. Suzuki

Writing, review, and/or revision of the manuscript: Y. Ito, H. Kimura, Y. Maeda, F. Ishida, K. Izutsu, Y. Isobe, M. Yamaguchi, K. Oshimi, R. Suzuki

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Ito, H. Kimura, J. Takizawa, S. Nakamura, K. Oshimi, R. Suzuki

Study supervision: K. Kawa, K. Oshimi, R. Suzuki

The authors thank staff at all participating institutions in this study: Tokyo Medical and Dental University, Kanagawa Cancer Center, Yokohama City University, Okayama University, Obihiro Kosei Hospital, Saga University, Yamanashi University, Shinshu University, NTT Medical Center Tokyo, Fukushima Prefectural Medical College, Kurashiki Central Hospital, Niigata University, Kyushu Medical Center, Nagano Red Cross Hospital, Tsukuba University, and Juntendo University.

The authors also thank the members of Central Pathology Review Board (Drs. Koichi Ohshima at Kurume University and Kengo Takeuchi at Cancer Institute), Central Imaging Review Board (Drs. Takao Kodama and Takanori Yano at Miyazaki University, and Yosuke Kakitsubata at Miyazaki Konan Hospital), and Data and Safety Monitoring Committee (Drs. Jin Takeuchi at Nihon University, Keizo Horibe at Nagoya Medical Center, and Keitaro Matsuo at Aichi Cancer Center) and Ms. Fumiyo Ando for excellent technical support for real-time quantitative PCR assay.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.