Clinical utility of plasma miR‐371a‐3p in germ cell tumors

Abstract Germ cell tumours predominantly of the testis ((T)GCTs) are remarkably chemotherapy sensitive. However, a small proportion of patients fail to be cured with cisplatin‐based combination chemotherapy. miR‐371a‐3p is a new liquid biopsy biomarker for (T)GCTs. The aim of this study was to evaluate clinical utility of plasma miR‐371a‐3p level in patients starting systemic chemotherapy. Patients were included before the first cycle (N = 180) and second cycle (N = 101) of systemic first line chemotherapy, treated between July 2010 and May 2017. Plasma miR‐371a‐3p levels were measured with the ampTSmiR test and compared to disease characteristics and outcome. Pretreatment plasma miR‐371a‐3p levels were increased in 51.7% of cases and associated with number of metastatic sites, presence of lung, retroperitoneal, and mediastinal lymph node metastases, S – stage, IGCCCG risk group, and response to therapy. Patients with a negative pretreatment plasma level had better progression‐free survival (PFS) and overall survival (OS) compared to patients being positive for miR‐371a‐3p (hazard ratio [HR] = 0.26, 95% confidence interval [CI] 0.09‐0.71, P = 0.02 for PFS and HR = 0.21, 95% CI 0.07‐0.67, P = 0.03 for OS, respectively). Patients negative for miR‐371a‐3p in both samples had a superior PFS (HR = 0.10, 95% CI 0.01‐21.49, P = 0.02) and OS (HR = 0.08, 95% CI 0.01‐27.81, P = 0.008) compared to patients with miR‐371a‐3p positive in both samples (multivariate analyses were non‐significant). In total 68% of the patients were S0. This study demonstrates clinical value of plasma miR‐371a‐3p level in chemotherapy naïve (T)GCT patients starting first line of chemotherapy to predict prognosis.

all malignancies in the age group between 20 and 40 years. 1,2 TGCTs represent a highly curable malignancy. In fact, cisplatin based chemotherapy is the mainstay in the treatment of TGCTs, resulting in about 70%-80% cure of patients with disseminated disease with a multimodality approach. However, between 20% and 30% of patients do not respond or relapse with this treatment protocol. 2,3 About 20%-25% of patients with relapsed TGCTs may be cured with various approaches, including salvage chemotherapy based on high-dose chemotherapy with autologous hematopoietic stem-cell rescue or combined with standard cisplatin dose and previously non-utilized drugs. [4][5][6] Currently, the clinical management of advanced TGCT disease is based on classical serum protein markers (alpha-fetoprotein [AFP], βhuman chorionic gonadotropin [HCG], and lactate dehydrogenase [LDH]), and imaging studies. 7,8 However, only 60% of TGCTs demonstrate elevated serum tumour markers at initial diagnosis. Therefore, in advanced cases, there is a need to identify reliable markers capable of determining chemotherapy effectiveness and patients' outcome. microRNAs (miRNAs) are involved in important biological processes, both normal as well as pathological, including various types of cancer.
The data showed an area under the curve (AUC) of 0.96, with a 90% sensitivity and 91% specificity. 14 In an independent study performed with a slightly different protocol and consisting of a series of sera from 166 TGCTs patients and 118 male controls, a positive miR-371a-3p level was found in 88.7% of the patients with the AUC of 0.94, and a 93.4% specificity. 10 Both studies demonstrated that the detection of miR-371a-3p in serum is far more sensitive and specific than the classical serum markers β-HCG, AFP, and LDH. Furthermore, the miR-371a-3p levels correlate with tumour burden and treatment results. 10 In addition, the four-serum embryonic stem cell related miRNA panel (including miR-371a-3p, miR-372-3p, miR-373-3p, and miR-367-3p) also showed a high sensitivity/specificity for diagnosing pediatric (mainly extra-gonadal) GCTs, based on serum and cerebrospinal fluid analyses. It allowed early detection of relapse of a single mixed tumour and distinguished intracranial malignant GCTs from intracranial non-GCTs at the time of diagnosis. 15 In addition, the ampTSmiR test targeting miR-371a-3p used in our laboratory is more sensitive than the conventional TGCTs protein serum biomarkers for the detection of residual disease and relapse. 13,14,18 In conclusion, miR-371a-3p is the most promising miRNA for the detection of (T)GCTs in body fluids to date, performed with a highly sensitive and specific multiplexed pre-amplification quantitative real-time PCR (qRT-PCR) technique. 10,14 However, in spite of the fact that miR-371a-3p is a promising new specific biomarker for (T)GCTs, data related to its clinical utility are still limited. [9][10][11][12]18

| Samples collection
Peripheral blood samples were collected from all translational study participants into BD Vacutainer ® EDTA tubes at baseline in the morning on day −1 or 0 of the first cycle of chemotherapy (n = 180) and before the second cycle of chemotherapy (n = 101). Patients' blood samples (5 mL) were centrifuged at 2300 g for 10 minutes to separate the plasma and blood cells. Collected plasma samples were afterwards filtered through a 0.2 μm filter to remove larger particles.
Plasma aliquots were stored at −80°C until further analysis. In total, 281 plasma samples of 180 patients were analysed (Table 1). Median age of patients was 30 years (range 16-67 years).
The miRNA levels of (T)GCT patients were compared with plasma levels of 50 male healthy donors, of which also serum was available taken at the same time point. Plasma was collected using Vacuette K3 EDTA Liquid, 6 ml vials (Greiner Bio-One, Alphen aan den Rijn, the Netherlands).
The median age was 54 (range 22-70 years). In our previously reported study, we showed that age did not influence the outcome of the miRNA analysis in sera 14

| miRNA purification and RT-qPCR
For the ampTSmiR tests specific miRNAs were isolated from 50 μL plasma using target-specific anti-miRNA magnetic beads as described before. 14 In short, cDNA generation and quantification of miRNA levels were performed with a highly sensitive multi-

| Quality control
A non-human miRNA spike-in ath-miR-159a was added in the same fixed amount to the sera (0.2 μL of a 1 nmol/L stock solution) for quality control of RNA isolation and cDNA generation. For calibration of input miRNA levels ath-miR-159a was used. All plasma samples were visually inspected and no hemolytic samples were present that could lead to false interpretation of the results. No samples had to be excluded due to poor miRNA recovery, based on recovery of the spike-in ath-mir-159a (variation in Ct generally within ± 2 Cts, n = 301, SD < 1.67). For normalization the mean levels of the endogenous reference miRNA (miR-30b-5p) were used as described before. 14,15 In each cDNA synthesis experiment, five 10-fold dilution series of purified miRNA of the TCam-2 seminoma cell line was included for quality control and qPCR efficiency and interplate calibration. For negative control, the no template control, elution buffer was added instead of purified miRNA as described previously. 14 T A B L E

| Evaluation
The target miRNA level per sample was determined according to 2−ΔΔCT method. 15 Threshold for miR-371a-3p was calculated using plasma samples of healthy donors analysed in this study ( Figure S1).
A cut-off value of ≥2.0 (Ct = 29.26) of the highest level (Ct = 30.26) observed in the healthy donor group was used to dichotomize the samples in two categories as positive or negative (cut-off for miR-371a-3p = 1) as described earlier. 15 Plasma samples of healthy donors (n = 50) contained similar minimal levels of miR-371a-3p as matched sera, and control sera used in a previous study (n = 104) 14 ( Figure S1).

| Statistical analysis
Patient data were tabulated. The patients' characteristics were summarized using the mean or median (range) for continuous variables and frequency (percentage) for categorical variables, respectively.
Statistical analysis was performed using non-parametric tests as the distribution of the miR-371a-3p levels was significantly different from the normal distribution (Shapiro-Wilk test). The Mann-Whitney U test was used for the analysis of the association of the miR-371a-3p expression to clinicopathological variables between the two groups of patients, and Kruskal-Wallis test among more than two groups, whereas Fisher's exact test or the χ 2 test were used when miR-371a-3p expression was categorized as positive or negative.
The associations between the histological subtypes of the primary (T)GCT and the plasma miR-371a-3p levels are summarized in Table S4. Within the whole group, there was no association between histology of primary tumour and plasma miR-371a-3p levels. In stage I disease, there were no differences in plasma miR-371a-3p levels based on histology of primary tumour.
Analysis of 101 paired samples (corresponding samples before starting of first cycle and second cycle of chemotherapy) showed that mean plasma level of miR-371a-3p was significantly higher before the first cycle compared to the second cycle of chemotherapy (mean ± SEM = 33.6 ± 8.1 vs. 0.2 ± 0.1, P < 0.00001) (Figure 1). Of 53 patients (52.5%) that were positive for pretreatment miR-371a-3p, only two patients (3.8%), remained positive, one with significant (from 99.9 to 1.5) and second with minor decrease (from 3.3 to 2.2) of plasma miR-371a-3p, while two patients (2.0%), initially miR-371a-3p negative, became positive before the second cycle of chemotherapy ( Figure S2A,B).

| Association between serum tumour markers and plasma miR-371a-3p levels
There was correlation between miR-371a-3p continuous levels and serum AFP and LDH, and between miR-371a-3p dichotomized and LDH (Table 3, Figure S3A). There was a moderate correlation between serum tumour markers defined as S-stage and plasma miR-371a-3p (Table 3, Figure S3B

| Prognostic value of plasma miR-371a-3p
In the median follow-up time of 20.9 months (range 0.  In multivariate analysis, pretreatment plasma levels of miR-371a-3p was not independently associated with IGCCCG risk group for PFS (P = 0.20) or OS (P = 0.33) in the whole population (Table 4) nor in the non-seminoma subgroup (not shown), while IGCCCG risk group category was associated with patients' outcome irrespective of plasma miR-371a-3p levels.  (Table 4, Figure 2A,B)  Plasma miR-371a-3p positivity was found in 13.3% of patients with clinical stage I disease after orchiectomy only. Surprisingly, there was no association between plasma miR-371a-3p levels and some prognostic factors associated with poor outcome, like non-pulmonary visceral metastases or extragonadal GCTs suggesting that miR-371a-3p is indeed a marker associated with tumour burden as suggested before, 10 also in line with the finding that this miR is expressed in all histological elements except teratoma, rather than with treatment resistance. Subgroup analysis revealed that this association is mainly triggered by non-seminomatous histology. However numerically, in seminoma similar trends were observed as in non-seminoma.
In our study we observed for the first-time prognostic value of pretreatment plasma levels of miR-371a-3p for PFS and OS in (T) GCTs. Subgroup analysis suggested limited prognostic value in seminoma patients and patients with negative serum tumour markers, e.g. subgroups for whom, the disease response could be evaluated only with imaging studies. However, extremely good prognosis of these subgroups and low number of events preclude to draw any definitive conclusion about the prognostic value of plasma miR-371a-3p in these subgroups. Multivariate analysis did not show prognostic value of plasma miR-371a-3p levels independently of IGCCCG risk category, an observation that is consistent with strong correlation between several IGCCCG risk factors and plasma miR-371a-3p level. Sharp decline in positivity of plasma miR-371a-3p after one cycle of cisplatin-based chemotherapy compared to serum tumour markers suggest rapid clearance of miR-371a-3p, in agreement with a recent study 10 and implies that plasma miR-371a-3p changes could be early, very sensitive predictor of treatment effect in TGCTs. 14,21 Plasma miR-371a-3p before the second cycle of chemotherapy showed no prognostic value in the whole patient population, probably due to rapid normalization of plasma miR-371a-3p after even one cycle of chemotherapy, an observation consistent with a previous study. 10 On the other hand, patients with positive miR-371a-3p in both samples had inferior outcome compared to continuously negative patients, suggesting that persistence of plasma miR-371a-3p positivity could be an early marker of treatment failure in TGCTs.
However, due to the low number of patients in subgroups, these results should be confirmed in future studies.
Despite several strengths, this study has some limitations as well, including the retrospective nature of the analysis. Even though this is currently the largest study evaluating plasma miR-371a-3p clinical utility in this patient population, low number of events due to very good prognosis of TGCTs preclude from drawing definitive conclusions in some subpopulation of patients. Moreover, due to the low number of events in subgroups for whom the clinical utility of plasma miR-371a-3p could be largest due to negativity of serum tumour markers (S0-stage disease and/or seminoma), we cannot assess the real clinical utility of plasma miR-371a-3p in these subgroups. In the study presented here the levels of miR-371a-3p have been investigated. We reported an AUC of 0.951, with an 89% specificity and 90% sensitivity for miR-371a-3p and an AUC of 0.962 for the combination of the miRs with specificity of 92% and sensitivity of 91%. 14 However, in our most recent studies addition of the extra miRs compared to miR-371a-3p did not improve the outcome. 13,18 In addition, other studies are based on only using miR-371. 10,16,21 In this large translational study we showed for the first time an association between plasma miR-371a-3p level and several disease characteristics including sites of metastases, serum tumour markers, IGCCCG prognostic group or favourable response to chemotherapy measured at the time just prior to the start of chemotherapy.
Moreover, we observed prognostic value of plasma miR-371a-3p in chemotherapy-naïve GCT patients starting first line of chemotherapy as well as prognostic value of plasma miR-371a-3p changes during the treatment suggesting clinical utility of plasma miR-371a-3p in TGCTs.