Detection of Recurrence through microRNA-371a-3p Serum Levels in a Follow-up of Stage I Testicular Germ Cell Tumors in the DRKS-00019223 Study

More than 90% of all patients with testicular germ cell tumors (GCT) can be cured with modern management (1). The majority of GCT cases present with clinical stage I (CSI) disease, and most of these patients are managed with surveillance after orchiectomy. The risk of recurrence in these cases depends on morphologic features of the primary tumor. In seminomas, approximately 20% of patients with primary tumor size of >4 cm will relapse, while smaller primaries involve gradually lower rates (2). In nonseminomas, lymphovascular invasion incurs a recurrence rate of 40% to 50%, while patients without this histologic feature will relapse in about 15% of cases (3). Follow-up (F/U) in CSI cases comprises cross-sectional abdominal imaging with CT or MRI, chest radiography, physical examination, and measurements of serum tumor markers alpha fetoprotein (AFP), beta human chorionic gonadotropin (bHCG), and lactate dehydrogenase (4). However, diagnosis of recurrence is hampered by low sensitivity of 37% to 60% of current imaging modalities for detecting lymph node metastases (5, 6). Only lymphadenopathies >1 cm may be characterized as metastases if located in the typical landing zones of testicular neoplasms (7). Similarly, serum tumor markers have a low sensitivity for detecting relapses (8, 9). bHCG is expressed in only 30% of all seminoma patients, but no more than 11% to 22% of relapses have elevations of this marker (9, 10). In nonseminoma, bHCG and/ or AFP are expressed in 70% of CSI cases (11), but relapse is detected by elevated markers in 41% to 61% of cases depending on the presence of lymphovascular invasion (12, 13). There is no expression of AFP in pure seminomas and no expression of either AFP or bHCG in pure teratomas.

In aggregate, the early detection of recurrences is considerably impeded by the low accuracy of currently available diagnostic modalities. Therefore, better tools for relapse detection are required. In recent years, evidence has accumulated suggesting the utility of serum levels of microRNA-371a-3p (M371) as a novel and highly sensitive serum biomarker for GCTs (14, 15). The sensitivity and specificity for diagnosing primary GCTs are reported to be 85% to 90.1%, and 89.1% to 99%, respectively (16–20). Notably, patients with recurrent GCT also revealed elevated M371 levels (21). However, the database regarding the sensitivity of M371 for detecting recurrence is still limited and there are also controversial results (22). Moreover, the expression rate in recurrences may be somewhat lower than that in primary GCT, as reported in one major series (19). However, other series reported equally high M371 sensitivities in both recurrences and in primary GCT (22–24).

In the current study, we prospectively evaluated the usefulness of the M371 test for detecting relapses in a series of patients with CSI GCT managed by active surveillance. In particular, the following three questions were addressed: (i) Is the M371 test capable of detecting any newly arising GCT disease during F/U of CSI patients under active surveillance? (ii) Can the test detect recurrence earlier than conventional methods? (iii) Do elevated postoperative M371 levels predict future recurrence?

Between June 2019 and November 2022, 352 patients with CSI GCT were prospectively enrolled from 23 urologic institutions in Germany, Austria, and Italy. Ninety-one patients were excluded for various reasons (Details in Fig. 1). The majority of excluded participants had provided less than three serum samples during F/U. Three patients had to be excluded because they had contralateral germ cell neoplasia in situ proven by biopsy during orchiectomy. Confounding of M371 measurements might have resulted in these cases because elevations may occur in 50% of these patients in the absence of metastases (25).

A total of 258 patients were eligible and managed by active surveillance (Table 1).

Follow-up visits involved imaging procedures, clinical examinations, and measurements of traditional tumor markers according to contemporary guidelines (refs. 4, 26; Supplementary Tables S1 and S2). In addition, serum levels of M371 were measured at each visit. Thus, in seminomas, M371 measurements were performed every six months, while in nonseminomas, the test was employed every three months during the first 2 years and six-monthly thereafter. A subsample of 64 patients also underwent M371 measurement prior to orchiectomy. The first measurement was performed within 2 weeks after orchiectomy.

Reference standard for relapse was an unequivocal and continuous radiographic enlargement of regional lymph nodes to >1.0 cm short axis diameter or rising tumor marker levels (AFP, bHCG) beyond the upper limit of norm (ULN) in the course of two consecutive visits or by both criteria. Patients who developed contralateral GCT were also rated as relapses. We included such secondary cancers because the aim of the current study was to analyze the ability of the test to diagnose any newly arising GCT lesions irrespective of their location. The median follow-up time was 18 months [interquartile range (IQR), 9–27 months]. Local caregivers of study patients were blinded to measurement results of the M371 test, thus no clinical decision-making was based on study results.

Ethical approval was provided by the Ethical committee of Ärztekammer Hamburg (MC 152/19, July 15, 2019) and by Ärztekammer Bremen (#301/17, September 21, 2017). The study was registered at Deutsches Register Klinische Studien (DRKS-00019223). Written informed consent was obtained from all patients. All study activities were conducted in accordance with the Declaration of Helsinki of the World Medical Association, amended by the 64th General Assembly in October 2013.

Whole blood samples were collected in 9 mL serum separation tubes (Sarstedt, Nümbrecht, Germany). Serum for the measurement of M371 expression was obtained after centrifugation and aliquots were stored in 2 mL cryotubes (Th Geyer, Renningen, Gemany) at −80°C until processing.

Serum M371 expression levels were measured as detailed earlier (19). Briefly, total RNA was isolated from 200 μL cubital vein serum using the miRNeasy Mini Kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions. Reverse transcription was performed for both miR-371a-3p and endogenous control miR-30b-5p using the M371 test (miRdetect, Bremerhaven, Germany). The miRNA expression was quantified on a 7500 Fast Real-Time PCR System (Applied Biosystems, Darmstadt, Germany) using the M371 test (miRdetect, Bremerhaven, Germany). The serum levels of M371 were measured relative to the endogenous control miR-30b-5p. The relative quantity (RQ) of miR-371a-3p was calculated using the ΔΔCt method (27). The measurements of classical tumor markers were performed in routine hospital laboratories according to institutional standard operating procedures.

Patient data regarding the histology of testicular tumors and age were registered in a commercially available database (MS Excel, version 2019) at study entry. During F/U, the results of the measurement of M371, classical tumor markers, and imaging procedures were added to the files at each time point of F/U. Statistical evaluation was performed using SPSS version 26 (IBM, Armonk, NY, RRID:SCR_002865). For comparison of continuous variables, the Mann–Whitney U test and Wilcoxon signed-rank test were used. Statistical significance was assumed at P < 0.05. The following performance characteristics were calculated with corresponding 95% confidence intervals (CI) by employing the highest measurement value obtained in each patient during F/U: Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and AUC of ROC curve. These characteristics were calculated for the entire cohort of patients with GCT and also separately for both seminomas and nonseminomas. Youden index analysis was used to determine the optimal cut-off value of serum M371 levels for identifying relapses.

On the basis of the currently available published data (22, 24), sensitivity and specificity of the M371 test both exceeding 90% and an AUC clearly above 0.9 would represent a positive result with respect to first goal of the study.

The complete original data set consisting of all raw data pertaining to this study is attached in the electronic supplement (Supplementary Table S3). All results of this study, all tables, and all figures are derived from this data set.

Clinical characteristics of the 258 patients enrolled in this study are listed in Table 1. Thirty-nine patients (15.1%) developed relapse at a median of 6 months (IQR, 3–12 months), two of whom developed contralateral GCT; all others had retroperitoneal lymphadenopathies. Of the relapses, 64% were detected by imaging techniques only, 26% by marker elevation only, and 10% by both modalities. The ROC curve revealed an AUC of 0.991 (95% CI, 0.982–1.000; Fig. 2). Youden index analysis revealed an RQ of 15 as optimal cutoff for differentiating relapses from healthy patients. Using this threshold, 47 patients were found to have elevated M371 levels during F/U, but only 39 of them had clinically confirmed relapse (Fig. 3A; Table 1). Eight of the 219 non-relapsing patients had to be considered false-positive results, while 211 were true-negative. If the cutoff of RQ = 5 as used in our former studies (19, 21, 25) had been applied, as many as 28 additional cases would have been rated false positive. The M371 test has a diagnostic sensitivity of 100% (95% CI, 100.0%–100.0%), and it has a specificity of 96.3% (95% CI, 93.9%–98.8%) for detecting relapses of CSI cases upon surveillance. Separate analysis of seminomas and nonseminomas revealed an AUC of 0.993 (95% CI, 0.984–1.000) and 0.985 (95% CI, 0.961–1.000), respectively (Supplementary Fig. S1A and S1B). Of note, the median M371 expression of false-positive cases was significantly lower (RQ = 37.6) than that of clinically confirmed recurrences (RQ = 153.7; P = 0.024). Table 2 provides details regarding the diagnostic performance characteristics of the test including PPV and NPV along with separate analyses of both seminomas and nonseminomas.

The classical markers bHCG and AFP attain much lower sensitivities for the detection of relapses than M371, but their specificities are in the same range as the M371 test (Table 2; Supplementary Fig. S2).

In 11 cases (28.2%), elevations of the M371 test preceded relapse detection with imaging techniques and/or marker elevations by 3 to 15 months. The time points of the first detection of relapse with conventional methods and the M371 test in each of the relapsing patients are listed in Supplementary Table S4. The Kaplan–Meier plot (Fig. 3B) reveals no significant difference between the median time to relapse detection with the M371 and with traditional methods (Log rank test P = 0.956). Both diagnostic modalities detect relapses at a median of 6 months, with no significant diagnostic advantage of M371 over conventional techniques.

Figure 4 illustrates the median M371 levels of relapsing patients and those without relapse at the time of orchiectomy, immediately after orchiectomy (median interval from surgery to miRNA test of 8 days), and at the last visit during follow-up. Serum levels at the last visit were much higher in relapsing patients than in non-relapsing (P < 0.001), as noted previously. However, both the preoperative and postoperative levels are not different among relapsing and non-relapsing patients (P = 0.52). Thus, the current data do not support the hypothesis that postoperative M371 levels can predict future relapse.

This study revealed three crucial results. First, there is clear evidence for the utility of the M371 test for detecting relapses in patients with CSI GCT during surveillance, featuring a sensitivity and a specificity of 100% and of 96.3%, respectively. Second, the test does not offer a significantly earlier relapse detection than imaging techniques and/or serum marker elevations. Third, elevation of M371 immediately after orchiectomy failed to predict the future risk of recurrence.

The M371 test has previously been shown to greatly outperform the classical tumor markers bHCG and AFP with respect to the primary diagnosis of GCTs (28, 29). The current study revealed evidence for another feature of the test: the ability to accurately detect relapses in CSI GCTs managed with surveillance. In fact, the diagnostic ability of the test appears to be more accurate in this setting than in the primary diagnosis of GCTs, as the sensitivity and specificity of 100% and 96.3%, respectively, as found in the current study, seemingly surpass the sensitivity and specificity of 90.3% and 94.1%, documented in the primary diagnosis setting (19). This small difference in diagnostic accuracy between the two clinical scenarios may relate to the low sensitivity of M371 in detecting small volume seminomas with less than 50% sensitivity in primary seminomas sized <1 cm. As the vast majority of relapsing cases in this study were confirmed radiologically with lymphadenopathies >1 cm, there were no particularly small tumor volumes among the relapsing seminoma cases in this study. Thus, the sensitivity of the test in this setting proved to be superior to the performance in the primary diagnosis, where usually around 10% of newly diagnosed seminomas are smaller than 1 cm (30).

A possibly low prevalence of teratomas among the relapsing cases might have indirectly contributed to the excellent performance of the test in this study. The insensitivity of the M371 test to detect teratomas is a well-documented weakness of the test (31), but this limitation obviously did not influence the overall results of this study because the number of teratomas was probably small. We hypothesize this, although we do not have histologic confirmation of relapsing nonseminomas. Teratomatous relapses typically tend to present late (32), but none of the relapses in the current study occurred later than 9 months after orchiectomy.

The separate analyses of seminomas and nonseminomas revealed widely equivalent performance characteristics of the test in both subgroups. Only PPV was lower in seminoma with 73.9% versus 91.7% in nonseminoma. Although the difference is still modest with widely overlapping 95% CIs of the two values, the disparity between the two subgroups does obviously relate to the higher number of false-positive findings in seminoma (n = 6) than in nonseminoma (n = 2) in conjunction with a higher number of seminomas (n = 189) than nonseminoma (n = 69) in the study population and the higher frequency of relapses in nonseminoma (32%) than in seminoma (9%).

Notably, in the current study, a cut-off value of RQ = 15 was used to identify relapses, which is somewhat higher than RQ = 5, the ULN currently used with the M371 test for detecting primary GCTs (19). Both threshold values were calculated using the Youden index analysis relating to the corresponding study populations; however, two reasons may account for the small difference regarding the cut-points, one clinical, and one technical. The patient population used for calculating the former cut-point consisted of patients with all clinical stages and a 58% proportion of seminomas from several European countries (19), whereas the current study population consisted of CSI patients only with a 73% proportion of seminoma from only Central European countries. Measurement of the M371 serum levels were performed with the commercially available M371 test kit (miRdetect, Bremerhaven, Germany) in this study. This test basically consists of the same technical steps as the formerly used method; however, due to minor refinements, it appears to detect its miRNA target with higher sensitivity than the former one. It is conceivable that the ULN of the M371 test should be adjusted to each clinical scenario in which the test is used. Such a methodology has recently been suggested for the assessment of post-chemotherapy residual masses of metastatic seminoma (33). However, for practical purposes, a uniform threshold value that fits all clinical scenarios is desirable. Such a uniform value would likely range around RQ = 10–12, but the ULN of M371 still needs to be determined.

Eight patients without relapse were found to have elevated M371 levels. Notably, the median M371 level found in false-positive cases was significantly lower than that of true-positives. Four cases were detected as M371 positive early during follow-up. They maintained mildly elevated levels at follow-up visits, but remained recurrence-free. Two other patients developed elevated M371 levels during later follow-up, but they have likewise remained recurrence-free ever since, although long-term follow-up is missing. Two others discontinued follow-up, and no further information was available. M371 elevations unrelated to GCTs have been reported in thyroid neoplasms (34), in COVID-19 infections (35), and in selected patients with testicular malignant lymphoma (36). However, it is unlikely that the false-positive patients in the current study were afflicted with such diseases. Thus, these eight cases had to be considered as false-positives, rendering the specificity of the test to 96.3% in the setting of relapse detection in patients with CSI GCT. As false-positive values may predominantly occur in elevations close to the ULN, it could be rational, practically, to repeat such measurements if no other evidence for relapse is given.

The excellent sensitivity of the M371 test to detect relapses in patients with CSI GCT during surveillance has been previously noted. The Toronto group reported a 94.1% sensitivity of M371 for detecting recurrences in 34 CSI patients using banked serum samples. In contrast to our study, the authors did not observe false-positive findings (22). A Swiss study reported 100% sensitivity of M371 in detecting relapses among 10 of 33 patients with CSI GCT prospectively followed. Notably, these authors also reported one false-positive finding (24). In a small Dutch case series, 3 of 3 patients with CSI GCT were shown to have elevated M371 levels at the time of relapse (37).

The traditional tumor markers bHCG and AFP revealed sensitivities of 35% and 25%, respectively, in the current study, and even the combined application of both markers yielded only a 45.2% sensitivity in the current study. These results are in accordance with previous reports that demonstrated the modest performance of traditional markers in relapse detection (8, 38). Series with high proportions of nonseminomas reported somewhat higher sensitivities of traditional markers of 40% to 73% (13, 39). Thus, the particularly low sensitivity of bHCG and AFP found in the current study may relate to the high proportion of seminomas in this study.

In aggregate, there is much evidence for the utility of the M371 test in detecting recurrences in early stage GCTs. However, the ability to detect relapses is not restricted to CSI cases because M371 is a universal marker for GCTs (40). Accordingly, elevated M371 levels have been reported in recurrences arising from various clinical scenarios other than surveillance in CSI (19, 20, 23, 41, 42).

Early case reports gave rise to the hope that M371 elevations may offer significantly earlier detection of relapses than conventional diagnostic techniques (37, 43). The current study revealed that earlier diagnosis of relapse with the test was in fact observed in 28% of patients. Specifically, in eleven cases, M371 elevations preceded relapse detection with imaging and/or marker elevation by 3 to 15 months. However, as shown by the Kaplan–Meier curve, the median time to relapse detection is 6 months with both M371 measurements and with standard technology. Thus, there is no statistically significant time gain with the M371 test until relapse detection. The Swiss study reported a median time-shift of two months with the M371 test until detection of relapses (24). The Toronto group found higher M371 levels at time points closer to the time of relapse detection. However, that result did not reach statistical significance (22). Overall, there might be a weak trend towards earlier diagnoses with the test. However, presently there is no clear evidence for clinically relevant antedating of relapse diagnoses with the M371 test.

With the advent of miRNA testing in GCTs, it has been speculated that persistent elevated M371 levels after orchiectomy could predict future relapses (19, 44). The current study clearly shows that M371 levels measured within 2 weeks after orchiectomy were not significantly different between patients destined for relapse and those staying healthy. This result is in line with the findings of both the Toronto group (22) and the Swiss study (24). Further support for the non-association of postoperative M371 levels with recurrences comes from the lack of any association between postoperative M371 levels and the presence of established risk factors for recurrence in nonseminomas (45). In all, the level of evidence for the insignificance of postorchiectomy M371 levels is sound, but recent data may contribute further pieces to the puzzle (46).

Eighty-two patients originally enrolled had to be excluded from study because they had provided less than 3 serum samples for analysis. This reduction of sample size may have reduced statistical power and may thus represent a possible weakness of the study. Also, timing of M371 testing deviated from F/U rosters with both additional and missed testings in about 40% of patients. Major reasons for this weakness were nonadherence of patients to F/U schedules and significant variability in adherence to recommended schedules. Non-compliance is a well-documented problem in the real world environment of surveillance of CSI patients (47, 48) and the current study was not free from it.

Another possible weakness of this study is the rather short median follow-up of 18 months. Although most relapses in patients with CSI GCT will arise within the first 2 years, some more may occur during the third year after surgery, and the latter were missed by the study. However, as the main goal of our study was to evaluate the diagnostic utility of the M371 test rather than analyze the frequency of relapses, the short follow-up period is probably a minor problem.

Lack of information regarding the clinical and histologic details of relapsing patients might be a shortcoming because patients with bulky disease could not be differentiated from those with low-volume tumor load.

A possible downside of the current study is that hemolysis of serum samples was examined macroscopically and not systematically using technical methods. Hemolysis may confound quantification of miRNAs in serum because high amounts of endogenous control miR-30b are released from damaged erythrocytes (49).

A methodologic weak-point could result from the fact that the cut-point of RQ = 15 used in this study had not been validated in an independent data set.

The strengths of the investigation relate to the prospective multicentric enrollment of patients and direct comparison of M371 with traditional markers in the majority of cases.

This study confirmed previous reports (22–24, 37) and thus significantly increased the level of evidence for the ability of the M371 test to detect relapses in patients with CSI GCT. Although some practical issues still need to be addressed such as validation of the cut-off level, the test now appears to be close to its implementation in follow-up schedules. Practically, M371 testing could be performed simultaneously with traditional tumor marker sampling. Positive test results are highly suggestive of the presence of active germ cell cancer, but clinical decision-making should presently be based on further diagnostic techniques, mainly imaging. Future research will show if the test alone is sufficient to accurately diagnose GCT relapses as suggested by the current data. One vision could be that part of the imaging procedures could be spared, thus reducing young subjects´ exposure to ionizing radiation, and even cost saving could come true.

G. Belge reports grants from Deutsche Krebshilfe during the conduct of the study; as well as reports ownership 8.3% of shares in miRdetect Company, Bremerhaven, Germany. J. Heinzelbecker reports grants from German Cancer Aid outside the submitted work. F. Zengerling reports personal fees from Janssen-Cilag GmbH, Pfizer Pharma GmbH, and Roche Pharma GmbH outside the submitted work. J. Frey reports personal fees from mirdetect GmbH during the conduct of the study; personal fees from mirdetect GmbH outside the submitted work; and is an employee of mirdetect GmbH. K.P. Dieckmann reports ownership 8.3% of shares of miRdetect Company, Bremerhaven, Germany. No disclosures were reported by the other authors.

G. Belge: Conceptualization, resources, formal analysis, supervision, funding acquisition, validation, investigation, writing–original draft, project administration, writing–review and editing. C. Dumlupinar: Formal analysis, investigation, writing–original draft, writing–review and editing. T. Nestler: Resources. M. Klemke: Formal analysis, investigation, writing–original draft, writing–review and editing. P. Törzsök: Resources. E. Trenti: Resources. R. Pichler: Resources. W. Loidl: Resources. Y. Che: Resources. A. Hiester: Resources. C. Matthies: Resources. M. Pichler: Resources. P. Paffenholz: Resources. L. Kluth: Resources. M. Wenzel: Resources. J. Sommer: Resources. J. Heinzelbecker: Resources. P. Schriefer: Resources. A. Winter: Resources. F. Zengerling: Resources. M.W. Kramer: Resources. M. Lengert: Investigation. J. Frey: Formal analysis. A. Heidenreich: Resources. C. Wülfing: Resources, supervision. A. Radtke: Conceptualization, formal analysis, writing–original draft, writing–review and editing. K.P. Dieckmann: Conceptualization, resources, formal analysis, supervision, validation, investigation, writing–original draft, project administration, writing–review and editing.

This study was supported by Deutsche Krebshilfe (No. 70113186/2019).

The authors thank the laboratory staff and nurses of the urologic institutions involved in the study for their technical assistance and logistic support. We thank all patients who participated in the above clinical study for their consent and participation.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.