Plasma thymus and activation‐regulated chemokine (TARC) as diagnostic marker in pediatric Hodgkin lymphoma

Abstract Pediatric classical Hodgkin's lymphoma (cHL) is characterized by Hodgkin Reed‐Sternberg cells located in an inflammatory microenvironment. Blood biomarkers result from active crosstalk between these cells. One promising biomarker in adult cHL patients is “thymus‐and‐activation‐regulated chemokine” (TARC). The objectives of this study were to define normal TARC values in non‐cHL children and to investigate and correlate pretherapy TARC as diagnostic marker in pediatric cHL. In this multicenter prospective study, plasma and serum samples were collected of newly diagnosed cHL patients before start of treatment (n = 49), and from randomly selected non‐cHL patients (n = 81). TARC levels were measured by enzyme‐linked immunosorbent assay. The non‐cHL patients had a median plasma TARC value of 71 pg/mL (range: 18‐762), compared to 14 619 pg/mL (range: 380‐73 174) in cHL patients (P < .001). TARC values had a high discriminatory power (AUC = .999; 95% confidence interval, .998‐1). A TARC cutoff level of 942 pg/mL maximized the sum of sensitivity (97.9%) and specificity (100%). TARC plasma levels were associated with age, treatment level, bulky disease, B‐symptoms, and erythrocyte sedimentation rate. TARC was found to be a highly specific and sensitive diagnostic marker for pediatric cHL. This noninvasive marker could be of great value as screening test in the work‐up for pediatric patients with lymphadenopathy.

diagnosis has been made, staging in pediatric cHL patients requires Fluor-Deoxyglucose-Positron Emission Tomography (FDG-PET) scans.
Although the latter is a sensitive test, it carries several disadvantages, including exposure to radiation, time consumption, high costs, and lack of specificity [5]. Blood biomarkers on the contrary are more easily available, cost-effective, and almost noninvasive for the patient. Therefore, they could be of great value to diagnose and stage pediatric cHL.
TARC is produced by HRS cells and antigen-presenting cells and attracts T-helper type 2 cells [9]. In adults, approximately 90% of cHL patients show positive TARC staining of HRS cells by immunohistochemistry and about 82-93% of patients have significantly elevated TARC levels in their pretreatment serum [10][11][12][13]. TARC is not expressed in adult patients with nonclassical nodular lymphocyte predominant Hodgkin lymphomas or non-Hodgkin lymphomas [14,15]. In adult cHL patients, higher TARC levels are correlated with presence of B-symptoms, bulky disease, metabolic tumor volume, and advanced disease stage [13]. Moreover, TARC levels decrease after one cycle of chemotherapy in both early and advanced stage cHL patients [6,13].
To our knowledge, TARC levels in pediatric patients with cHL have not been reported. Besides, normal values of TARC are not yet determined in children and may differ from those in adults. Because pediatric and adult cHL differ partly with respect to histology, risk factors, clinical presentation, and staging at diagnosis and prognosis [16][17][18], further research is warranted to determine whether TARC is a diagnostic biomarker in pediatric cHL patients as well. Therefore, the aim of this study is to determine normal TARC values in children without cHL (non-cHL) and to investigate whether TARC can be used as a diagnostic marker in pediatric cHL patients, using biopsy results as gold standard.
Moreover, we study the association of TARC values with stage of disease, presence of B symptoms, erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), and bulky disease. Therefore, these diseases were checked and recorded [20][21][22].

Patient inclusion
This study was IRB-approved, and registered under Dutch Trial registry number 6876.

Plasma and serum collection
Recently, Zhao et al showed that TARC levels are significantly lower in platelet-free plasma samples than in platelet containing serum samples, because platelets contain TARC. [23,24]. To investigate whether plasma and serum are both a useful source of TARC and to compare their reliability as diagnostic markers, we collected serum and plasma samples for each participant in the study.

TARC detection
After collection, serum tubes were stored at room temperature in a vertical position prior to centrifugation. Plasma tubes were centrifuged within 30 min after arrival at the laboratory. Blood sample centrifugation was performed at room temperature for 15 min at 1000 × g. Afterward serum and plasma supernatant was collected and  leave-one-out cross validation (LOOCV) was implemented. The optimal threshold was determined based on Youden's index, which maximizes the sum of sensitivity and specificity. Before study start, it was obtained that with at least 40 cHL patients available, a 90% two-sided confidence interval for sensitivity with lower limit 85% would be produced when the sample sensitivity is 95% (analogously for non-cHL controls and specificity).

Statistical analyses
To compare TARC plasma and serum AUCs, DeLong's method for paired data AUCs was used.
All statistical tests were two-sided. We used R (version 3.4.1) and SAS (version 9.4) for analyses.  However, our study population had a significantly different distribution of stages with less patients in stage I and II, and more patients in stage III and IV (P = .012).

Patient's characteristics
The median age of the non-cHL group was 13 (range: 1-17) years.

Normal TARC values in non-cHL group
Non-cHL patients had a median TARC value of 71 (range: 18-762) pg/mL for plasma and 317 (range: 27-1300) pg/mL for serum. TARC levels were associated with age: for 1-year increase in age, a 4% decrease in TARC plasma was found, and a 4% decrease for TARC serum. We did not detect elevated TARC plasma and serum levels in the eight cases of atopic dermatitis in this group (Table S1).

TARC levels are highly elevated in the blood of pediatric cHL patients
The median pretreatment plasma TARC level in 47 cHL patients was 14 619 pg/mL (range: 380-73 174). All but one patient had elevated TARC levels (97.8%).  No difference was found in the AUCs for the correlated ROC curves of paired plasma and serum data (P = .413).

TARC levels correlate with parameters of tumor burden
We assessed the association between TARC levels and disease characteristics (Figure 3 and Tables S2 and S3). Patients with bulky disease had significantly higher plasma and serum TARC levels compared to patients without (approximately threefold increase for plasma and serum). Plasma and serum TARC levels were significantly associated with the presence of B-symptoms at diagnosis (91% and 40% increase for plasma and serum, respectively) and ESR > 30 mm/h (approximately three-and twofold increase for plasma and serum, respectively). Patients with a higher treatment level and patients with organ involvement and E-lesions had significantly higher plasma but not serum TARC levels (slightly less than twofold increase or higher).
No significant associations were found with CRP or stage. The fact that TARC is associated with treatment level but not with stage F I G U R E 3 TARC levels are correlated with disease burden. A, Patients with bulky disease had significantly higher plasma and serum TARC levels compared to patients without bulky disease (215% and 192% increase for plasma and serum, respectively). B, Plasma and serum TARC levels were significantly associated with the presence of B-symptoms at diagnosis (91% and 40% increase for plasma and serum, respectively). C, Plasma and serum TARC levels were associated with ESR > 30 mm/h (215% and 95% increase for plasma and serum, respectively). D, Patients with a higher treatment level and patients with organ involvement and E-lesions had significantly higher plasma TARC levels (at least 190% increase), but this was not the case for serum TARC * Other: The patient with relapsed HL was treated with a relapsed treatment protocol where treatment levels were not used.
could be due to the strong correlation of TARC with bulky disease, Bsymptoms, and ESR, characteristics that are used to determine treatment level. When introducing these three variables together with stage in a multiple regression model, bulky disease remains significantly associated with TARC plasma and serum levels (slightly less than threefold increase or higher in TARC for presence of bulky disease vs not).

DISCUSSION
This is the first study to report normal TARC values in children. We define a cutoff level for TARC elevation in children with cHL. Our data demonstrate that TARC level in plasma or serum is a sensitive and specific diagnostic marker for pediatric cHL, with histology as gold Consistent with adult studies, TARC levels in pediatric cHL patients correlated with bulky disease [10,11,13] at diagnosis, and with B-symptoms and ESR [11,13,25]. A correlation with disease stage was not found unlike results from most studies in adults [11,13,26].
We detect a trend toward higher levels of TARC in stage 3 and 4 disease, compared to stage 2 disease, although this is not significant.
There were no patients with stage 1 disease and low patient numbers with stage 2 disease included in this study. This may explain why we did not find a correlation of TARC levels with disease stage. However, we did find a correlation between TARC and treatment levels according to the EuroNet-PHL-C2 protocol, which is a derivative of staging that takes into account systemic signs of inflammation such as B-symptoms, ESR level, bulky disease, and presence of E-lesions.
The fact that TARC is associated with treatment level but not with stage could be due to the strong correlation of TARC with bulky disease.
TARC levels in pediatric non-cHL patients closely correspond to TARC levels of adult non-cHL patients investigated in other studies [11,23]. TARC levels found in the non-cHL group are very consistent and stay in a comparable range. We therefore think that these patients form a representative non-cHL group. In our non-cHL group, TARC plasma and serum levels decreased with age; further research is necessary to explain this.
In contrast to earlier studies, where children and adult patients with atopic dermatitis and eczema had elevated levels of TARC [20][21][22][23]; this was not the case in the eight non-cHL patients with atopic dermatitis included in our study.
TARC is measurable in both plasma and serum, although normal values and cutoff levels are different. No difference was found in the AUCs for the correlated ROC curves of paired plasma and serum data. Therefore, both serum and plasma can be used to measure TARC as diagnostic marker.
In this study, plasma and serum samples were all taken before start of treatment with chemotherapy. A possible limitation of this study is the fact that in the majority of cases, a complete lymph node was removed for pathological diagnosis before the TARC plasma and serum samples were drawn. In patients with lower tumor burden and stage, this might have influenced the amount of TARC, because TARC is produced by HRS cells present in the lymph nodes.
We propose that blood TARC levels can be of great value as screening test in the diagnostic work-up for pediatric patients with lymphadenopathy. Especially cHL is sometimes difficult to differentiate from other pathologies due to the slow growing pattern and, sometimes, the lack of specific symptoms. Diagnostic markers such as ESR are useful but nonspecific in this work-up. Based on this study, elevated TARC has high sensitivity and specificity for cHL. Further research in patients with other causes of lymphadenopathy is necessary to define how TARC should be implemented in the diagnostic algorithm.
In conclusion, our study shows that TARC is a valuable diagnostic biomarker for children with cHL, with a sensitivity of 97.9% and specificity of 100% for plasma (cutoff level of 942 pg/mL) and 100% sensitivity and 100% specificity for serum (cutoff level of 2257 pg/mL). TARC levels are correlated with bulky disease, treatment level, the presence of B-symptoms, and ESR.

ACKNOWLEDGMENTS
We thank all patients for donating their samples. This work was financially supported by the Erasmus MC Foundation, enabled by a legacy of the family Etienne-van Dijk. We would like to thank them both.

AUTHOR CONTRIBUTIONS
AB and FM-W conceived the project and provided leadership. EZ, RH, WW, FM-W, and AB organized the study. WW collected the non-cHL group samples. EZ, RH, and ML analyzed the data and contributed to the manuscript. ML made the figures. EZ wrote the manuscript. FM-W and AB supervised the study. All authors reviewed and accepted the contents of the article.