Utility of serum tumor markers during surveillance for stage I seminoma
The serum tumor markers α-fetoprotein (AFP), β-human chorionic gonadotropin (HCG), and lactate dehydrogenase (LDH) are often measured as part of surveillance protocols in patients with stage I seminoma. In this study, the authors evaluated the utility of routine measurement of these markers in the detection of disease relapse.
Data were gathered from a prospectively maintained database of patients who underwent surveillance for stage I testicular seminoma diagnosed between 1982 and 2005 at Princess Margaret Hospital. Patients were followed on a predefined schedule with physical examination (PE), serum tumor markers, abdominopelvic computed tomography, and chest x-rays. The records of patients who relapsed were examined for details of imaging and serum tumor markers throughout the period of follow-up until the time of relapse.
Of the 527 patients who were managed by surveillance, 75 patients (14%) relapsed at a median follow-up of 72 months. Of these, 65 patients relapsed within the first 3 years and had routine serum tumor markers measured. In total, 11 patients had abnormal tumor markers at the time of relapse (AFP, 0 patients; HCG, 6 patients; LDH, 4 patients; and HCG and LDH, 1 patient). Only 1 patient had an elevated tumor marker (LDH) before relapse, as defined by an abnormal imaging study (n = 64) or physical examination (n = 1), for which the treatment and outcome were not affected.
Serum tumor marker levels did not aid in the early diagnosis of disease relapse in patients with stage I seminoma who were managed with surveillance. The current results indicated that routine measurement of serum tumor markers can be discontinued safely in seminoma surveillance schedules. Cancer 2012. © 2012 American Cancer Society.
Surveillance is the management option of choice in patients with stage I seminoma.1-3 Although surveillance schedules may differ, most patients are followed with chest x-rays (CXR), computed tomography (CT) scans of the abdomen and pelvis (CT-AP), and measurement of the serum tumor markers α-fetoprotein (AFP), β-human chorionic gonadotropin (HCG), and lactate dehydrogenase (LDH).4, 5 It has been demonstrated that knowledge of serum tumor markers levels is useful in the diagnosis and prognostication of germ cell tumors (GCTs),1, 6 and studies have examined the role of tumor markers in detecting GCT relapse in patients with nonseminomatous GCTs (NSGCTs) or in heterogeneous populations of patients with GCTs.5, 7-9 However, it is unclear whether routine measurement during surveillance in patients with stage I seminoma has any value in the early detection of relapse in addition to regular imaging studies.10, 11
The focus of more recent work has been to optimize follow-up schedules to minimize the “burden” that men face after a diagnosis of stage I seminoma.5, 11 However, the optimal surveillance schedule has not been established. Thus, as a part of this optimization process, we recently published a report on the lack of utility of CXR in seminoma surveillance.11 In the current study, we continue this work and examine the role of routine measurement of serum tumor markers in the identification of relapse in patients with stage I seminoma who undergo surveillance.
MATERIALS AND METHODS
Data were gathered from a prospectively maintained institutional database after we obtained local research ethics board approval. The medical records of all patients with histologically confirmed, stage I testicular seminoma who were managed with surveillance between 1982 and 2005 were examined to allow for a minimum of 3 years of follow-up. Patients initially were managed with surveillance as part of a phase 2 study in the 1980s; then, from the early 1990s, it was recommended to patients as the preferred option. Although the follow-up schedule had evolved over the years examined, as depicted in Table 1, most patients were followed with CXR and CT-AP every 4 to 6 months for 7 years and annually thereafter.12 For median follow-up calculations, patients who were discharged after 10 to 15 years of surveillance and those who were lost to follow-up were censored at that point. In the majority of patients, serum tumor marker levels (AFP, HCG, and LDH) were obtained at each follow-up visit at least every 4 months for the first 3 years and in the early part of the period examined up to the early 1990s, more frequently. The diagnosis of relapse was made on clinical grounds, such as abnormal imaging studies and/or physical examination. Confirmatory pathology from biopsy also was obtained wherever feasible up to the early 1990s. Data were abstracted on all patients who relapsed, including all imaging studies and serum marker values between the time of diagnosis and the time of relapse. At relapse, patients routinely were fully restaged with CXR and chest CT, CT-AP, and repeat serum markers for patients who developed a relapse beyond 3 years. Normal serum markers were identified according to the normal laboratory values at the time of relapse with the exception of LDH, for which a level >1.5 times the upper limit of normal was considered abnormal (as denoted in Table 2).
Table 1. Typical Surveillance Schedule for Patients With Stage I Seminoma at Princess Margaret Hospital During the Study Period
|1||—||—||—||Markers, CT-AP||—||—||—||Markers, CT-AP, CXR||—||—||—||Markers, CT-AP|
|2||—||—||—||Markers, CT-AP, CXR||—||—||—||Markers, CT-AP||—||—||—||Markers, CT-AP, CXR|
|3||—||—||—||Markers, CT-AP||—||—||—||Markers, CT-AP||—||—||—||Markers, CT-AP, CXR|
Table 2. Abnormal Serum Tumor Markers in Patients Who Relapsed During the Stage I Seminoma Surveillance Protocol
Of the 527 patients who were managed by surveillance between 1982 and 2005, 519 patients were alive at the time of data analysis, 1 patient had died of seminoma, and 7 others had died of unrelated causes at last follow-up. In total, 176 patients in the examined cohort were lost to follow-up, including 19 who were lost within 2 years of diagnosis and 157 who were lost within 5 years of diagnosis. At a median follow-up of 72 months (range, 1-193 months), 75 of 527 patients (14%) had relapsed. In 74 patients (98.6%), the relapse was detected by a demonstration of abnormal lymphadenopathy on CT-AP. All CT-APs were performed according to our standard surveillance protocol (ie, none of the patients experienced symptoms that led to an earlier CT-AP). In 1 patient, relapse was detected by an abnormal physical examination, and biopsy-proven, asymptomatic inguinal lymph node metastasis was identified. The median time to relapse was 13.4 months (range 1.7-108 months). Relapse was predominantly in the para-aortic region, with 64 of 75 relapses detected in lymph node(s) isolated to this region. In 6 patients, relapse was detected in the pelvic lymph nodes (common/external/internal iliac) in the absence of enlarged para-aortic adenopathy, whereas 3 patients relapsed with enlarged adenopathy in both the para-aortic and pelvic regions. Two patients relapsed with isolated inguinal adenopathy.
In total, 65 patients relapsed in the first 3 years after diagnosis and, thus, had routine tumor markers evaluated. Of these 65 patients, 11 patients had abnormal markers at the time of relapse. Details of the tumor marker values, patterns of relapse, and disease burden at relapse in these patients are listed in Table 2. No patient had an abnormal AFP level, 6 patients had elevated HCG alone, 4 patients had elevated LDH alone, and 1 patient had abnormalities in both HCG and LDH. None of the patients who had abnormal tumor markers at relapse had abnormal values at the time of initial diagnosis; however, because some patients underwent orchiectomy at external institutions, preorchiectomy LDH values were not available for 2 of the 5 patients who had abnormal LDH values at relapse. However, in these patients, the LDH level was normal at the time of entry into the surveillance program.
Because patients were followed in an organized seminoma surveillance clinic, there was good compliance with follow-up and blood work. Of the 7 patients who had an abnormal HCG level at the time of relapse, only 1 patient had a minor protocol violation, with tumor markers evaluated 5 months before relapse (instead of 3-4 months, because the relapse was within 1 year). Of the 5 patients who had abnormal LDH levels, 2 did not have blood drawn within 6 months of the relapse date, because they had blood drawn at 7 months and 9 months before relapse. Of the 54 patients who relapsed with normal marker levels within 3 years of diagnosis (during which tumor markers were examined routinely), 4 patients had not had tumor marker levels measured within 6 months of the date of relapse (in those 4 patients, the time from previous tumor marker measurements to relapse was 6 months, 7 months, 9 months, and 13 months).
No relapse was diagnosed based on abnormal markers alone, because even the patients who had elevated markers had their relapse detected by CT scans at that time. None of the patients who had an abnormal tumor marker level at relapse had an abnormal value at the preceding surveillance visit, with the exception of a single patient who had an elevated LDH level of 744 U/L at the visit preceding his relapse detection but had a normal CT scan at that time. The elevated level was noted, and the patient continued with the standard surveillance protocol until relapse was detected on a CT-AP study 4 months later, at which time his LDH was 2114 U/L.
There were 10 patients who relapsed beyond 3 years and, thus, had not been followed with routine tumor marker levels. All relapses in this group were detected by an abnormal CT-AP. None of these patients had abnormal tumor markers at diagnosis or on restaging at the time of relapse.
Preorchiectomy HCG measurements were not available for 69 of the 527 patients who were followed on surveillance. Of the remaining 458 patients who had preorchiectomy serum tumor marker levels available, 39 patients had abnormal preorchiectomy HCG levels, which normalized postoperatively. Although 5 of 39 patients relapsed, none had abnormal levels of HCG at the time of relapse. Similarly, 12 patients had abnormal LDH levels before orchiectomy, but none of those patients experienced a relapse. Seven patients had a spurious rise in their tumor marker values at some point between diagnosis and relapse. Two patients had a rise in HCG (to 9.9 IU/L and 8.4 IU/L), 1 patient had a rise in AFP (17 μg/L), and 4 patients had a rise in LDH (range, 333-738 U/L). The tumor marker levels in all of these patients returned to normal before the detection of a relapse. Two patients who had temporarily elevated LDH levels (333 U/L and 346 U/L) had abnormal HCG levels at the time of relapse. The elevated LDH level in both of these patients was measured within 4 weeks of orchidectomy; thus, it was unclear whether the LDH elevation was related to disease.
To our knowledge this is the first report to specifically examine the utility of tumor markers in detecting relapse in patients with stage I seminoma who were followed on a surveillance policy. All 527 patients were followed with CT-AP studies at regular intervals; and, although 11 of 65 relapses that were observed in the first 3 years occurred among patients who had an abnormal tumor marker at the time of relapse, the abnormal marker was not the initial indicator in any of the observed relapses. There was 1 patient who had an abnormal LDH level 4 months before the identification of relapse on a CT scan. Although the abnormal value was noted at the time, given the negative imaging study, a sole elevated LDH level was not enough to be considered indicative of relapse or to be used to direct treatment. In those who relapsed subsequent to the period during which routine serum tumor markers were measured (ie, >3 years), the routine measurement of serum tumor marker levels would not have contributed to the detection of relapse. The data presented suggest that the routine measurement of serum tumor markers does not allow the earlier detection of disease relapse in patients with stage I seminoma who undergo surveillance compared with other measures, such as imaging studies and physical examination, and thus may be safely omitted in surveillance follow-up schedules. The frequency of imaging studies used in our surveillance schedule should be noted when interpreting these results; however, even when this is considered, the sensitivity of serum tumor markers as a diagnostic test of relapse is poor and unreliable.
Previous reports have suggested that measurement of LDH levels is of little benefit for the detection of relapse in patients who are treated for GCTs,7, 9 although these earlier studies also included patients with NSGCT. Trigo et al observed that an elevated tumor marker was the only initial evidence of recurrent disease in 40% of patients with NSGCT and in 25% of patients with seminoma.8 However, their retrospective study included a heterogeneous population, with some receiving adjuvant treatment and others followed on a surveillance schedule. In addition, the imaging was not standardized, because patients were followed with CXR and ultrasound or with CT-AP (because CT use became more common after 1981). It is not clear whether the proportion of seminoma relapses detected with elevated tumor markers alone was secondary to infrequent and/or less sophisticated imaging. Consequently, the results from that previous study are not relevant enough to direct the schedule of follow-up for patients with stage I seminoma who participate in contemporary surveillance programs.
Similarly, Mead et al13 recently reported the patterns of relapse in 2466 patients with stage I seminoma from 3 randomized trials who received adjuvant radiation or carboplatin. Those authors demonstrated that, of the 98 relapses reported, only 2 were “marker only” relapses. Although those data support the suggestion that routine tumor markers are of minimal, if any, value in stage I seminoma, because this was a population of patients who received adjuvant therapy, its relevance to surveillance protocols is unclear. Aparicio et al14 reported on 314 patients with stage I seminoma, of whom 100 were followed on a surveillance protocol. The authors reported that 36.4% of relapses in their cohort were detected by an increase in HCG (54.5% of relapses were detected through routine scans, and 9.1% were detected by symptoms of disease). Unfortunately, the report on that study did not indicate the method of relapse detection in the patients who were followed on surveillance, although, because only 6 relapses were observed in that surveillance group, it would be hard to apply any results to clinical practice.
The recent American Society of Clinical Oncology Clinical Practice Guidelines on the use of serum tumor markers in men with GCTs indicate that such markers continue to be useful in the diagnosis and prognosis of GCTs,10 although they probably are more useful in NSGCTs than in pure seminomas.15-18 Historically, serum tumor markers may have had value in the detection of nonseminomas that were misclassified on pathology, but this may no longer apply, because diagnostic techniques have advanced significantly in pathology. Modern immunohistochemistry using CD117 (mast/stem cell growth factor receptor) and D240 (a novel mesothelial marker) has resulted in the main improvements in diagnostic accuracy for pure seminoma. Previously, placental alkaline phosphatase, which also is positive in other GCTs, was the only immunostain available; however, CD117 and D240 are negative in NSGCTs.19
Although venipuncture is minimally invasive and is associated with minor risk, routine measurement of tumor markers when there is no advantage in detecting a relapse potentially may cause harm, particularly in the presence of a false-positive result, which occurred in several patients in our cohort. There are several noncancer-related conditions that lead to elevated tumor marker levels.9, 20-22 For example, AFP often is elevated in liver ailments (cirrhosis, hepatitis, alcohol abuse, etc) and has been reported in patients who were being followed for GCT.20, 21 The cost savings from eliminating routine tumor markers would be relatively modest; for instance, when current, direct laboratory assay costs were included in the model reported by Warde et al,23 the removal of routine blood work for the first 3 years would save $268 (in Canadian dollars) per patient. Although admittedly small, such savings further decrease the excess cost that surveillance has over adjuvant radiotherapy (not factoring in the cost of second malignancies that may be associated with adjuvant radiotherapy).
In a subset of patients with pure seminoma, there is a mild HCG elevation because of the presence of syncytiotrophoblastic giant cells.24, 25 The degree of HCG elevation reportedly is directly proportional to tumor burden,24, 26 although there are case reports of patients with markedly elevated HCG and stage I disease.24, 26, 27 The prognosis for patients with seminoma who have mildly elevated HCG at diagnosis is unclear, because it is uncertain whether it heralds an increased risk of relapse compared the risk for patients with normal levels who are undergoing surveillance.28-30 It is possible that a proportion of patients who presented with elevated HCG levels before orchiectomy had received adjuvant radiotherapy and, thus, would not have been followed in our seminoma surveillance program. Although the small number of patients (n = 5) in our series who initially had an elevated HCG level before orchiectomy and ultimately developed a relapse did not have an elevated HCG level at the time of relapse, we recommend gathering additional data to determine the role of routine measurement of tumor markers in detecting relapse in this subset of patients.
The value of tumor marker measurement at the time of relapse is not addressed in this study. It is noteworthy that 5 of the 11 patients who had a relapse with abnormal serum tumor marker levels in our cohort had a relatively large tumor burden (measured as a lymph node mass >5 cm in greatest dimension). Unfortunately, given the nature of our historic database and lack of access to CT-AP images for a significant proportion of patients because they relapsed before the advent of computerized imaging, we are unable to provide more detailed review of the relapses. As tumor markers (and particularly LDH) may be surrogates of overall tumor burden, measurement at the time of relapse should be continued, because it may influence treatment decisions.
This study has several limitations. The first is that, although our database was maintained prospectively, the study was a retrospective chart review of patients who presented and were managed over a 23-year period and, thus, is subject to potential association bias. In addition, as described above, the follow-up schedule and the normal limits for laboratory values changed over the study period. The number of patients who experienced a relapse was relatively small. Another issue that needs to be considered when interpreting our results is the proportion of patients who were lost to follow-up. Because seminoma affects young men at the most mobile periods of their lives, it is not surprising that a significant proportion of our patients (157 patients) had <5 years of follow-up. Concern over the possibility that these men may have relapsed after they were lost to follow-up, thus, may have affected the results, although this possibility is minimized by the finding that only 19 men had <2 years of follow-up. Because most relapses occurred within this time, and because all patients had routine tumor marker measurements recorded according to our protocol, we do not expect that there were enough relapses in this group to affect our conclusions.
In conclusion, the data presented suggest that, with the assumption of availability of expert pathologic diagnosis to rule out NSGCT elements, routine measurement of serum tumor markers provides no additional value to CT-AP studies and physical examinations for detecting relapse during surveillance for stage I seminoma. After initial staging, serum tumor markers may be safely omitted from stage I seminoma surveillance protocols.
No specific funding was disclosed.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.