Serum level of soluble interleukin‐2 receptor is positively correlated with metabolic tumor volume on 18F‐FDG PET/CT in newly diagnosed patients with diffuse large B‐cell lymphoma

Abstract Diffuse large B‐cell lymphoma (DLBCL) is the most frequent subtype of non‐Hodgkin lymphoma. High total metabolic tumor volume (TMTV) calculated using 18F‐FDG PET/CT images at diagnosis predicts poor prognosis of patients with DLBCL. However, high cost and poor access to the imaging facilities hamper wider use of 18F‐FDG PET/CT. In order to explore a surrogate marker for TMTV, we evaluated the correlation between the serum levels of soluble interleukin‐2 receptor (sIL‐2R) and TMTV in 64 patients with DLBCL, and the results were verified in an independent validation cohort of 86 patients. Serum levels of sIL‐2R were significantly correlated with TMTV. ROC analysis revealed that the cutoff value of TMTV ≥150 cm3 or sIL‐2R ≥ 1300 U/mL could predict failure to achieve EFS24 with areas under the curve (AUC) 0.706 and 0.758, respectively. Each of TMTV ≥150 cm3 and sIL‐2R ≥1300 U/mL was significantly associated with worse 5‐year overall survival and event‐free survival. Importantly, each of sIL‐2R <1300 U/mL or TMTV <150 cm3 identified patients with favorable prognosis among NCCN‐IPI high‐intermediate and high‐risk group. Serum level of sIL‐2R represents a convenient surrogate marker to estimate metabolic tumor burden measured by 18F‐FDG PET/CT that can predict treatment outcomes of patients with DLBCL.


| INTRODUCTION
Diffuse large B-cell lymphoma (DLBCL) is the most common subtype of malignant lymphoma, accounting for 30%-40% of non-Hodgkin lymphoma. 1 Although rituximab combined with cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) has led to a remarkable improvement in the treatment of DLBCL patients, 2 considerable proportion of the patients fail to cure. 3 To date, many prognostic factors are advocated, including patient factors: age, performance status (PS), 3 Charlson Comorbidity Index, 4 tumor burden; LDH, 5 clinical stage, 5 and biological features of tumor cells; and germinal center B-cell (GCB) type or non-GCB type, 6,7 CD5-positivity, 8 status of Epstein-Barr virus, 9 and double-hit lymphoma. 10 These factors should be considered comprehensively at diagnosis to estimate the prognosis of the patients with DLBCL. 18 F-FDG PET/CT is widely performed at initial staging in DLBCL patients. High total metabolic tumor volume (TMTV) calculated using 18 F-FDG PET/CT images at diagnosis is predictive of poor prognosis of DLBCL, follicular lymphoma (FL), and peripheral T-cell lymphoma. [11][12][13][14] However, not all patients can undergo 18 F-FDG PET/CT due to various reasons, including high cost, poor access to the imaging facilities, and urgent requirement for treatment initiation, encouraging us to explore a surrogate marker for metabolic tumor volumes. Several previous studies have shown that serum level of sIL-2R is a prognostic biomarker of DLBCL. 15,16 In the current study, we retrospectively evaluated the correlation between the serum levels of sIL-2R and TMTV based on 18 F-FDG PET/CT images at diagnosis and compared the role of these parameters as prognostic biomarker in newly diagnosed DLBCL.  19 with exclusion of physiological accumulation including urinary, myocardial, and brain FDG uptake. Nodular or heterogeneous uptake in the bone marrow was included as tumor involvement based on radiologist's interpretation, while diffuse uptake was considered as physiological uptake. All quantitative parameters were retrospectively measured by a nuclear medicine physician (ET) in a blinded fashion.

| Statistical analysis
Overall survival (OS) was calculated from the day of diagnosis until death or last follow-up. Event-free survival (EFS) was defined as time from diagnosis to disease progression, relapse after response, death, or last follow-up. The probabilities of OS and EFS were estimated using a Kaplan-Meier method, and differences between patient groups were analyzed using the log-rank test. The baseline patient characteristics were tabulated to check imbalance in the demographic information. and TMTV was assessed by using Pearson's product-moment correlation coefficient, respectively. All P-values were 2-sided, and a P-value of 0.05 was used as the cutoff for statistical significance. All the statistical analyses were performed with the EZR (http://www.jichi.ac.jp/saitamasct/SaitamaHP.files/statmedEN.html). 22 3 | RESULTS

| Patient characteristics
Baseline patient characteristics were listed in Table 1. In the training cohort, the median patient age at diagnosis was 74 years, ranging from 33 to 86 years. PS was 2 or greater in 30% of the patients, and 81% of the patients had stage III or IV. Sixty-six percent of the patients had extranodal involvement, including bone marrow (17%), and 58% presented with B-symptoms. Sixty-seven percent of the patients had elevated serum LDH value than normal level. For NCCN-IPI scores, 24%, 31%, and 45% of the patients were classified as Low or Low-intermediate (Low-int) risk group, High-intermediate (High-int) risk group, and High-risk group, respectively.
Subgroup analyses included the patients with NCCN-IPI High-Int and High (n = 49) demonstrated that the cutoff value of TMTV 150 cm 3 stratified treatment outcomes | 957 SENJO Et al.

| Validation of the results in the validation cohort
Finally, the prognostic impacts of serum levels of sIL-2R and TMTV, and correlation between sIL-2R and metabolic parameter were validated in the independent validation cohort composed of significantly younger patients with better PS, less advanced-stage disease, and lower NCCN-IPI scores than the training cohort ( Table 1). The OS and EFS in this cohort were shown ( Figure S2). In terms of TMTV and sIL-2R, there were no significant differences between patients in the training cohort and the validation cohort (Table 1). Kaplan-Meier curves showed that OS and EFS rates in patients with TMTV ≥150 cm 3 were again lower than in those with TMTV <150 cm 3 (5-year OS; 87.0% vs 59.5%, P = 0.016, 5-year EFS; 72.8% vs 52.3%, P = 0.0154; Figure S3A, B).
The median serum sIL-2R level at diagnosis was 1274 U/mL, ranging from 200 to 39 798 U/mL. Kaplan-Meier curves showed that sIL-2R ≥1300 U/mL was a strong prognostic factor both for worse OS and EFS (5-year OS; 86.3% vs 61.8%, P = 0.0188, 5-year EFS; 85.0% vs 46.8%, P = 0.000413; Figure S3C, D). Pearson's correlation tests gave similar results that there were positive correlations between sIL-2R and TMTV (R 2 = 0.461; P = 0.00000631; Figure 2B). In a univariate analysis, TMTV was associated with poor 5-year OS, whereas sIL-2R and TMTV were identified as poor prognostic factors for EFS (Table 2). In a multivariate analysis including sIL-2R, age was an independent prognostic factor and there was a strong trend toward worse 5-year OS in patients with higher sIL-2R (Table 3). In another multivariate analysis including TMTV showed that age, LDH, and TMTV were independent prognostic factor for 5-year OS (Table 3). Altogether, we could validate that both sIL-2R and TMTV are promising prognostic biomarkers and there is a positive correlation between sIL-2R and TMTV, suggesting that sIL-2R is useful for extrapolation of TMTV.

| DISCUSSION
Soluble IL-2R is a soluble form of the α-subunit of high-affinity receptor for IL-2 that consists of three subunits: α-subunit, β-subunit, and γ-subunit. While resting lymphocytes, monocytes, and NK cells constitutively express the β-and γ-subunits, the α -subunit of IL-2R (IL-2Rα) is constitutively expressed only on the cell surface of lymphoid neoplastic cells and transiently induced on the activated normal lymphocytes. 25 Although the precise mechanism of sIL-2R shedding is not clear, it has been shown that the release of sIL-2R is proportional to its cell surface expression, suggesting that serum levels of sIL-2R represent the numbers of IL-2R α-expressing lymphoma cells and activated lymphocytes. 26 Previous studies have shown the predictive role of pretreatment TMTV and sIL-2R for survival in patients with FL 13,27,28 and DLBCL. 11,15,16 Ennishi et al reported that sIL-2R > 1000 mg/dL predicted worse OS and EFS after R-CHOP in patients with newly diagnosed DLBCL. Goto et al reported that sIL-2R ≥1300 U/mL predicted worse prognosis both in GCB-like and in non-GCB-like DLBCL classified based on Hans criteria. 29 The cutoff value of sIL-2R in the current study was slightly different from those reported in previous studies. 15,16 Cutoff value of sIL-2R in our study was determined using CLIA, while ELISA was used in Goto's report. Although it is possible that the different methods for sIL-2R assessment could result in the different cutoff values, it could be also possible that the different endpoints used in ROC analyses performed by Goto   We have extended these previous findings on the prognostic values of TMTV and serum levels of sIL-2R at diagnosis. The primary aim of this study was to clarify the correlation between serum levels of sIL-2R and TMTV and compare the role of these factors as prognostic biomarkers. We found significant positive correlation between serum levels of sIL-2R and TMTV. Furthermore, sIL-2R ≥1300 U/mL stratified patients with poor prognosis in an analogous manner to TMTV ≥150 cm 3 did. These cutoff values also improved risk stratification of patients with NCCN-IPI High and High-Int.
Serum levels of sIL-2R have been routinely measured in DLBCL patients since 1990s in Japan. 30 This biomarker is also used as a major prognostic biomarker for transplant-related GVHD in United States. 31 Even though we have 4.3 PET units per 1 million people in Japan (3rd in the world), some hematology/oncology centers are not equipped with PET, and patients need to travel to the external PET facilities (OECD stat. 2017 https://stats.oecd.org/index.aspx?DataSet-Code=HEALTH_STAT#). Instead, we can know the serum levels of sIL-2R within a day of blood sampling in many centers. In Japan, the costs associated with 18 FDG-PET and sIL-2R are ~1000 and 40 USD, respectively. Thus, serum level of sIL-2R is a promising biomarker that can be easily and inexpensively measured in clinical practice and have a great potential as a predictor of outcome in DLBCL patients; however, it should be noted that PET/CT is also useful for monitoring of tumor responses after treatment, suggesting that these two tests could work together in a complementary fashion. It was reported that profound reduction of TMTV from initial PET/CT to interim PET-CT was associated with better prognosis, indicating that TMTV on interim PET/CT could be the useful biomarker in DLBCL. 32 Further studies are required to determine whether serum levels of sIL-2R after treatment could be correlated with TMTV on interim PET/CT. The accurate evaluation of tumor burden at diagnosis became more important in the rituximab era, because high tumor amounts promote clearance of rituximab from the circulation both in mice and in humans; higher TMTV at diagnosis of DLBCL led to lower rituximab exposure and inferior OS and PFS, 33 suggesting that rituximab dosing could be guided by tumor amount at diagnosis. Metavol ® is a free and open-source software tool to measure TMTV from PET/ CT scans. Although this software made measuring TMTV much easier, our data indicate that sIL-2R also correlates with tumor burden and enables us to evaluate tumor burden in patients, in whom PET/CT evaluation is not available. It should be noted that serum level of sIL-2R cannot function as all the same to PET/CT scan does, such as visualization the distribution of the tumor lesions in patients.
Our study has some limitations, including a retrospective setting, small sample size, rather low AUC in ROC analyses, diagnosis according to the previous 2008 WHO classification, lack of central review for pathological diagnosis, and use of R-CHOP-like chemotherapies in some patients. However, OS rate of 53.1% and EFS 45.4% at 5 years were consistent with previous studies, in which aged DLBCL patients were treated with similar regimens used in our study. 23,24,34,35 The difference in sIL-2R measurement between the training and validation cohorts might impact our results; CLIA was used in the training cohort, while ELISA was used in the validation cohort. Although the upper normal limits of sIL-2R in these two assays were similar (496 U/mL for CLIA and 500 U/mL for ELISA), it might be possible that serum sIL-2R levels differ slightly with methods of measurement kits. At least, the cutoff value of sIL-2R determined in the training cohort successfully stratified the outcome of the patients in the validation cohort. The correlation between CLIA-based and ELISA-based levels of sIL-2R needs to be clarified in the future studies. Another limitation of the current study is that the cameras and equipments used for PET/CT imaging were different from those used in the validation cohort.
In summary, we for the first time showed positive correlation between the serum level of sIL-2R and the quantitative parameter TMTV in patients with newly diagnosed DLBCL. sIL-2R is easily measurable in the clinical practice and have a great potential to predict treatment outcomes and assess metabolic tumor burden of DLBCL patients.