Usefulness of Serum Soluble Interleukin-2 Receptor Levels for Differentiating between PCNSL and SCNSL

Abstract

Serum soluble interleukin-2 receptor (sIL-2R) is a practical tumor marker that is elevated in hematogenous tumors. The purpose of this study was to determine the usefulness of serum sIL-2R for differentiating among malignant brain tumors, including primary central nervous system lymphoma (PCNSL) and secondary central nervous system lymphoma (SCNSL). This study retrospectively investigated the sIL-2R levels in 130 patients with various types of malignant brain tumors, including PCNSL patients (n = 48) and SCNSL (n = 8); metastatic brain tumors (MTs, n = 16); and glioblastoma (GBM, n = 58). The median sIL-2R level (U/mL) of the PCNSL, SCNSL, MTs, and GBM groups were 489.7, 1024.8, 413.3, and 332.7 respectively. The sIL-2R level was significantly higher in the SCNSL group than in the PCNSL or other groups. The area under the ROC curve generated from the sIL-2R level was 0.826 (sensitivity: 0.875, specificity: 0.667, cutoff value: 521 U/mL) for differentiating SCNSL from PCNSL and 0.685 (sensitivity: 0.667, specificity: 0.707, cutoff value: 342 U/mL) for differentiating PCNSL from GBM. Measurement of sIL-2R level was convenient and useful to differentiate between SCNSL and PCSNL, both of which demand different treatment strategies.

Introduction

For malignant brain tumors, accurate histologic diagnosis has been essential, because treatment depends on the tumor type.^1-8) Moreover, the surgical procedure, such as removal or biopsy, is determined by the preoperative diagnosis; therefore, preoperative misdiagnosis results in inappropriate surgery. Although making a diagnosis by imaging alone can be sometimes difficult, because malignant brain tumors often have resembling imaging features, radiologic images, such as computed tomography (CT) scan and magnetic resonance (MR) imaging, had been helpful for preoperative diagnosis. As previously reported, ring enhancements on contrast-enhanced T1-weighted MR imaging are the common findings in glioblastoma (GBM) and metastatic brain tumors (MTs)^9-11) and that the apparent diffusion coefficient (ADC) values on MR imaging were effective in differentiating between GBM and primary central nervous system malignant lymphoma (PCNSL).^12,13) However, because they both show similar MR imaging findings of homogeneously contrast-enhanced intraparenchymal lesions,^14-16) PCNSL is often difficult to distinguish from secondary central nervous system malignant lymphoma (SCNSL).

In addition to imaging, serum tumor markers are routinely used for the diagnosis of systemic cancers. However, some cerebrospinal fluid or serum markers are less effective for brain tumor diagnosis because of the influence of the blood-brain barrier (BBB).^17,18) Serum soluble interleukin-2 receptor (sIL-2R) is thought to be released from the surface of tumor cells and had been utilized as a diagnostic marker or scale of degree in hematologic malignancies.^19-21) In addition, although as best as we know, no report has investigated the use of serum sIL-2R levels for the differential diagnosis of malignant brain tumors, including PCNSL and SCNSL, we aimed to retrospectively examine and determine the clinical usefulness of the serum sIL-2R levels in patients with suspected malignant brain tumors, as elevated levels of serum sIL-2R have been detected in infections, autoimmune or other inflammatory diseases, and nonlymphoid and lymphoid malignancies.^22,23)

Materials and Methods

Patient and data selection

This retrospective study was conducted with the approval of the ethics committee of the Yamagata University School of Medicine. Informed consent in written form or optout announcement was obtained from all patients and/or relevant persons for this study. We retrospectively reviewed the medical records and radiologic features of cases with histologically confirmed malignant brain tumors and were treated at our institution from December 2007 to December 2019.

Serum sIL-2R measurement

Preoperative serum sIL-2R level was measured with a sandwich-enzyme immunoassay (BML; Tokyo, Japan). The normal range was defined as 122-496 U/mL, following the manufacturer's information. In this study period, we measured sIL-2R as a routine preoperative examination of patients with suspected malignant brain tumor.

Radiologic evaluations

All patients were examined by 1.5 or 3 Tesla MR imaging scanners (Achieva 1.5 or 3.0T, Philips Medical Systems, Amsterdam, Netherlands). Diffusion-weighted imaging was acquired with b-values of 0 and 1000 s/mm² and was used for generating ADC maps. For data evaluation, the Picture Archiving and Communication System (Impax ee, Agfa Healthcare, Bonn, Germany) was used. We set up several 50-mm² regions of interests (ROIs) in the contrast-enhanced lesion on the T1-weighted images of each tumor. We examined the minimum ADC (min ADC) value of the corresponding ROI on the ADC maps.

Histopathologic evaluations

Tumor specimens from biopsy or removal of brain tumors were available in all patients. We made a pathologic diagnosis of GBM in accordance with the WHO Classification 2016.¹⁾ The pathologic diagnoses of MTs, SCNSLs, and PSCNL were made by proficient pathologists.

Statistical analysis

Comparisons of age, sIL-2R level, and min ADC value among all groups were made using Kruskal-Wallis test. Furthermore, differences between the two groups were evaluated using Mann-Whitney U-test, followed by posthoc group comparisons with Bonferroni test. Comparisons of the proportions of multiple lesions and of lesions showing ring enhancements were made using Chi-square test. All data were considered statistically significant if the P-values were <0.05.

We evaluated the sensitivity (Sn) and specificity (Sp) of the serum sIL-2R level in differentiating SCNSL from PCNSL and PCNSL from GBM. We evaluated the ROC curve of the min ADC values for discriminating between PCNSL and GBM, with ROC curves being used to determine the diagnostic accuracy and cutoff values of the serum sIL-2R level.

The analyses were performed using EZR version 2.4-0 (Saitama Medical Center, Jichi Medical University), which is a graphical user interface for R (The R Foundation for Statistical Computing).²⁴⁾

Results

A total of 130 patients (48 PCNSL, 8 SCNSL, 16 MTs, and 58 GBM) were analyzed. The serum sIL-2R levels and other parameters of each group are shown in Table 1. The median age of the patients was 68 years (range, 26-87 years). The proportion of cases with multiple lesions was significantly higher in the PCNSL and SCNSL groups, and the proportion of cases with ring enhancement was significantly higher in the MTs and GBM groups than in the other groups. Representative cases that were difficult to diagnose by MR imaging are shown in Fig. 1. The mean sIL-2R levels significantly differed and were 489.7 U/mL (range, 164-2,380 U/mL); 1024 U/mL (range, 408-2720 U/mL); 413.3 U/mL (range, 253-848 U/mL); and 332.7 U/mL (range, 100-2,280 U/mL) in the PCNSL, SCNSL, MTs, and GBM groups, respectively. The mean sIL-2R level was significantly higher in the SCNSL group than in other groups (vs. PCNSL: P = 0.014, vs. MTs: P = 0.009, and vs. GBM: P < 0.001). In addition, the sIL-2R levels of both the PCNSL and MTs groups were higher, compared with that of the GBM group (P = 0.007 and P = 0.027, respectively) (Fig. 2a). The mean min ADC value was significantly higher in the GBM group than in the PCNSL (P = 0.009) and SCNSL groups (P = 0.018); there were no significant differences among the other groups (Fig. 2b).

Table 1.

Population characteristics

	All	PCNSL (n=48)	SCNSL (n=8)	MTs (n=16)	GBM (n=58)	P
Age (median & range)	68	68 (26-87)	66 (40-78)	68 (39-79)	67.5 (50-85)	0.44*
sIL-2R (U/mL) (mean & range)	443.2	489.7 (164-2380)	1024.8 (408-2720)	413.3 (253-848)	332.7 (100-2280)	<0.001*
multiple lesions (%)	36.9	58.3	75	43.7	12.1	<0.001**
ring enhancement (%)	92.3	18.7	0	81.3	100	<0.001**
min ADC value (10−6 mm3) (mean & range)	664.1	616.9 (369-900)	538.9 (437-713)	670.5 (253-1168)	718.6 (214-1287)	0.001*

PCNSL: Primary central nervous system lymphoma, SCNSL: Secondary central nervous system lymphoma, MTs: Metastatic brain tumors, GBM: Glioblastoma, sIL-2R: serum soluble interleukin-2 receptor, ADC: apparent diffusion coefficient

P-value*: comparison results of all groups using Kruskal-Wallis test.

P-value**: comparison results of all groups using Chi-square test.

Fig. 1.

Contrast-enhanced T1-weighted MRI findings in patients with malignant brain tumor.

The lesion shows ring enhancement in PCNSL (a) and as almost homogeneous contrast-enhanced lesions in SCNSL (b), MT (c), and GBM (d).

Fig. 2.

Box and whisker plots of the serum sIL-2R levels (a) and minimum ADC values (b) of the PCNSL, SCNSL, MTs, and GBM groups are shown.

ROC curve analysis

The ROC curve generated from the sIL-2R levels for differentiating SCNSL from PCNSL is shown in Fig. 3a. The area under the curve (AUC) was 0.826. The curve was nearest to the left corner of the plot when the cutoff value was 521 U/mL, at which point the sensitivity (Sn) and specificity (Sp) were 0.875 and 0.667, respectively. Similarly, the ROC curve for differentiating PCNSL from GBM is shown in Fig. 3b. The AUC was 0.685, and the curve was nearest to the left corner of the plot when the cutoff value was 342 U/mL, at which point the Sn and Sp were 0.667 and 0.707, respectively. The ROC curve generated from the min ADC values for differentiating PCNSL from GBM is shown in Fig. 3c. The AUC was 0.68, and the curve was nearest to the left corner of the plot when the cutoff value was 662 × 10⁻⁶ mm³, at which point the Sn and Sp were 0.688 and 0.621, respectively.

Fig. 3.

The ROC curves generated from serum sIL-2R levels for differentiating SCNSL from PCNSL (a) and PCNSL from GBM are shown (b). The ROC curve generated from the minimum ADC values for differentiating PCNSL from GBM is shown (c).

Discussion

In this study, we confirmed that MR imaging parameters, such as contrast-enhanced T1-weighted images and minimum ADC value, were useful for the differential diagnosis of malignant brain tumors, where, although differentiating among the malignant brain tumors is sometimes difficult when the MR imaging does not show the typical imaging findings, some neuroradiographic characteristics for these differential diagnosis of malignant brain tumors have been mentioned in several studies. In particular, the use of MR imaging alone was difficult to discriminate between SCNSL and PCSNL, both of which have different treatment strategies; therefore, a rigorous diagnostic workup is required. Despite the usefulness of whole-body examinations for differentiating between PCNSL and SCNSL, Mohile et al. reported that of 57 cases, 6 (11%) cases of SCNSL could not be diagnosed by whole-body CT scan alone, but all were diagnosed by ¹⁸F-Fluorodeoxyglucose (FDG) positron emission tomography (PET).²⁵⁾ SCNSL is often diagnosed by a combination of systemic CT scan and bone marrow biopsy. However, FDG-PET is not available at all facilities. Therefore, we considered whether serum sIL-2R can be useful for the differential diagnosis of malignant brain tumors, being as it is, measured easily, in combination with other examinations.

The mechanism and functional significance of elevated serum sIL-2R level for each pathologic condition remain unclear. In recent years, Murakami et al. used ROC curves to show the usefulness of serum sIL-2R in differentiating systemic lymphoma from other diseases.²¹⁾ In addition, there were some studies that compared systemic lymphoma with other diseases.^20,22,23) For PCNSL, Kitai et al. reported that serum sIL-2R levels were significantly higher in PCNSL than in intracranial diseases, which were difficult to distinguish from PCNSL by imaging studies and physical examination alone.¹⁹⁾ In this study, in addition to confirming the available reports, we first reported the usefulness of serum sIL-2R in distinguishing SCNSL from PCNSL. There had been no reports that explained the higher serum sIL-2R level in SCNSL than in PCNSL. We speculated that the difference in serum sIL-2R levels between PCNSL and SCNSL depended on the involvement of the BBB, which may prevent passage and leakage into the blood. We suggest that the possibility of SCNSL needs to be considered and investigated by whole-body examinations, such as FDG-PET in cases with high serum sIL-2R levels.

Moreover, we investigated the usefulness of serum sIL-2R in the differentiation between PCNSL and GBM, which is often difficult by imaging findings alone. As a result, q significant increase in serum sIL-2R level was observed in the PCNSL group. Although the min ADC value was showed by some reports to be useful for discriminating PCNSL from GBM,^12,13,26) its effectiveness was almost the same as that of serum sIL-2R in this study. However, the sIL-2R cutoff value compared was slightly higher than or within the normal range; therefore, the use of sIL-2R alone is difficult to make a diagnosis. We considered that serum sIL-2R, in combination with neuroradiographic features, such as multiple lesions, ring enhancement, and min ADC values, may be a reference marker for differentiating between PCNSL and GBM. The absence of significant difference in the sIL-2R level between the PCNSL and MTs groups was considered to be because of the fact that serum sIL-2R levels were found to be elevated even in patients with solid body tumors.^22,23) Although a systemic CT scan or bone marrow biopsy is required for the diagnosis, clinical staging, and treatment strategy of SCNSL, evaluation of sIL-2R in combination with other diagnostic imaging features may contribute to differentiating between SCNSL and PCNSL.

The present study had some limitations. The number of patients was limited; more patients are required for a better understanding of the utility of sIL-2R. Nevertheless, we considered it important to demonstrate that sIL-2R was a useful marker for differential diagnosis, especially between SCNSL and PCNSL.

Conclusions

For the first time, we reported that the serum sIL-2R level was significantly higher in SCNSL in PCNSL. In practice, serum sIL-2R may not be very useful in differentiating PCNSL from GBM or MTs. However, it could be highly useful in the differentiation between PCNSL and SCNSL, which are often difficult to distinguish and require totally different treatment strategies.

Conflicts of Interest Disclosure

The authors have no conflicts of interest related with this study.