Fax: +011-44 (208) 643 8809
Late recurrence in 1263 men with testicular germ cell tumors†
Multivariate analysis of risk factors and implications for management
Article first published online: 23 JUL 2002
Copyright © 2002 American Cancer Society
Volume 95, Issue 3, pages 520–530, 1 August 2002
How to Cite
Shahidi, M., Norman, A. R., Dearnaley, D. P., Nicholls, J., Horwich, A. and Huddart, R. A. (2002), Late recurrence in 1263 men with testicular germ cell tumors. Cancer, 95: 520–530. doi: 10.1002/cncr.10691
This work was undertaken in The Royal Marsden NHS Trust, which received a proportion of its funding from the NHS Executive. The views expressed in this publication are those of the authors and not necessarily those of the NHS Executive.
- Issue published online: 23 JUL 2002
- Article first published online: 23 JUL 2002
- Manuscript Accepted: 19 FEB 2002
- Manuscript Received: 23 JAN 2002
- Institute of Cancer Research
- Bob Champion Cancer Trust
- Cancer Research Campaign
- testicular carcinoma;
- germ cell tumor;
- late recurrence;
Testicular germ cell tumors are highly curable. However, 10–30% of patients have recurrence after initial treatment. The time–course of recurrence has implications for the duration of follow-up. This study was undertaken to assess the risk and time–course of recurrence and to identify patients at higher risk of late recurrence.
The records of 1263 patients with primary testicular germ cell tumors presenting to the Royal Marsden Hospital between December 1979 and December 1993 were reviewed. In all, 255 episodes of recurrence were documented (including 44 patients with multiple recurrences) and used to calculate recurrence-free survivals.
Fifty-three patients (15 seminomas; 38 nonseminomatous germ cell tumors [NSGCT]) had recurrence more than 2 years after initial presentation. A multivariate analysis of risk of recurrence after 2 years identified positive markers at presentation and the presence of differentiated teratomas in postchemotherapy surgical specimens as significant predictors. Very late recurrence (> 5 years) occurred mainly in patients with metastatic NSGCT (12 of 14 patients) with a 1% annual risk of recurrence between 5 and 10 years. Very late recurrence was also seen in one case of metastatic seminoma and one case of Stage I NSGCT managed by surveillance. Most late recurrences (n = 9) were detected at routine annual follow-up visits but five had recurrences with symptoms leading to an unscheduled clinic visit.
Late recurrences are rare in patients with testicular germ cell tumors and follow-up to detect recurrence may not be needed after 5 years, except in those presenting with metastatic NSGCTs. Cancer 2002;95:520–30. © 2002 American Cancer Society.
Testicular germ cell tumors are highly curable. However, 10–30% of patients have a recurrence after initial treatment. The majority of recurrences happen in the first 2 years after treatment.1, 2 Recurrences after this are rare and their time–course and pattern have implications for long-term follow-up. Currently, there is no consensus on the long-term follow-up policy of these patients. Some authors have suggested life-long follow-up for all patients with testicular carcinoma due to the very small possibility of late recurrence1, 3 (arbitrarily defined as recurrences after 2 years). Others have suggested discharge from routine oncologic care after 5 years.2
This study was designed and performed to document the rate and time–course of recurrence in patients with primary testicular germ cell tumors. We describe the pattern of late recurrence with the aim of identifying patients at higher risk and we report the method of detection. The implication of these findings on follow-up policy is discussed.
MATERIALS AND METHODS
All patients with primary testicular germ cell tumors presenting initially to the Academic Unit of Radiotherapy and Oncology at the Royal Marsden NHS Trust from December 1979 to December 1993 were analyzed. Patients who had their initial treatment elsewhere were excluded from our analysis to ensure a valid denominator for recurrence rates. Informed consent was obtained from patients for their respective management. Patients and recurrences were identified from the Bob Champion Unit database. These data were collected prospectively. Time from baseline at the date of diagnosis to first recurrence was measured. For subsequent recurrences, the time of previous recurrence was taken as the baseline date. Therefore, all patients were included in our at-risk denominator. Patients were grouped according to their presenting pathology (seminomatous vs. nonseminomatous germ cell tumors [NSGCT]) and stage of disease (Stage I vs. Stage II vs. Stages III and IV). Staging was performed according to the Royal Marsden staging system.4 During the study period, patients were followed up according to the schedule outlined in Table 1. Recurrence-free survivals were calculated according to the Kaplan–Meier method for each group.
|Stage I surveillance||2||2||3||4||12|
|Stage I radiotherapy||3||4||6||6||12|
|Stage I surveillance||1||2||3||6||12|
For patients with late recurrence (> 5 years after last treatment), the clinical, laboratory, and radiologic features at presentation and at recurrence were examined in detail. Hospital files were reviewed to identify how late recurrences were detected.
The estimation of lifetime risk of recurrence after 2 or 5 years is technically challenging. Because the data are right censored, with varying periods of follow-up, the denominator in such an estimate diminishes with time, thus making the estimate of risk of recurrence at a particular time point difficult. The Kaplan–Meier method of analyzing these results may be the most practical approach to gain an overall account of these data. To account for the decreasing denominator with increasing follow-up, the risk of recurrence after 2 and 5 years was calculated using the number of patients at risk at 5 and 10 years as the denominator, respectively. This gives the maximum risk in these time periods and may overestimate the risk.
Multivariate stepwise logistic regression analysis was used to identify factors predictive of recurrence after 2 years. An obvious limitation of this analysis is that some very late recurrences will not have yet occurred.
Factors that were considered for univariate and multivariate analysis were age at baseline, presenting histology (seminoma vs. nonseminoma), presence of differentiated teratoma in initial specimens, stage and substage, serum tumor markers at presentation (positive vs. negative), presence of differentiated teratoma in postchemotherapy excision specimens, response to treatment (complete response with negative markers vs. no complete response), and number of previous recurrences.
A total of 1263 previously untreated patients with primary testicular germ cell tumors and initial presentation to the Royal Marsden Hospital were treated between December 1979 and December 1993. Characteristics of patients at presentation are summarized in Table 2. In all, 255 episodes of recurrence were documented in 211 patients (including 44 second recurrences). Median follow-up for the group was 10.2 years. Median time to recurrence was 10.5 months (1.3–169 months). Recurrence-free survivals for seminoma and nonseminoma patients according to stage are presented in Figures 1 and 2. Table 3 summarizes the data for each group.
|Nonseminoma (all cases)||731|
|With TD elements||392|
|Positive at baseline||766|
|Negative at baseline||500|
|No marker at baseline||208|
|No of previous recurrences|
|Patients at baseline||No. of recurrences||Months to recurrence||RFS (%) 95% confidence interval (No. at risk)||Recurrences after 2 yrs||Recurrences after 5 yrs|
|2 yrs||5 yrs||10 yrs|
|Stage III, IV||60||13||3–20||76.8||76.8||76.8||0||0|
|Stage III, IV||293||73||2–169||79.4||75.2||72.6||19||9|
Prediction of Recurrence
Fifty-three patients (15 seminomas; 38 NSGCTs) had recurrence of disease more than 2 years after initial presentation. To identify risk factors for recurrence after 2 years, stepwise logistic regression analysis was performed. Results of univariate and multivariate analysis for all patients are presented in Table 4.
|Hazard ratio||Lower 95% CI||Upper 95% CI||P value||Hazard ratio||Lower 95% CI||Upper 95% CI||P value|
|Age at baseline|
|Presence of TD|
|NSGCT with TD||392||1.268||.662||2.430||.474||.996|
|NSGCT without TD||492||1|
|Serum tumor marker|
|Not marker positive||708||1||1|
|Response to treatment|
|CR marker negative||1294||1|
|Posttreatment excision results|
|TD absent in excised specimen||92||.722||.171||3.038||.656||.665||.158||2.808||.579|
|TD present in excised specimen||111||4.304||2.219||8.348||.000||3.420||1.720||6.801||.000|
|No. of previous recurrences|
Initial stage, i.e., the presence of metastatic disease at presentation, positive tumor markers at presentation, and the presence of differentiated teratomas in postchemotherapy surgical specimens were predictive of recurrence in univariate analysis. Presenting positive tumor markers and the presence of differentiated teratomas in postchemotherapy specimens remained statistically significant in multivariate analysis (hazard ratio: 1.9, P = 0.035 and 3.4, P < 0.001, respectively).
The same analysis was done for patients with NSGCTs (Table 5). Identical factors (stage, markers, and presence of differentiated teratoma elements in postchemotherapy surgical specimens) were predictive of late recurrence with statistical significance in univariate analysis. In multivariate analysis, only the presence of teratoma in postchemotherapy surgical specimens remained significant (hazard ratio: 3.9, P < 0.001).
|Hazard ratio||Lower 95% CI||Upper 95% CI||P value||Hazard ratio||Lower 95% CI||Upper 95% CI||P value|
|Age at baseline|
|Serum tumor markers|
|Not marker positive||289||1|
|Response to treatment|
|CR marker negative||777||1|
|Posttreatment excision results|
|TD absent in excised specimen||81||.738||.171||3.187||.684||.738||.171||3.187||.684|
|TD present in excised specimen||110||3.904||1.914||7.961||.000||3.904||1.914||7.961||.000|
|No of previous recurrences|
Recurrence Pattern in Seminoma Patients
Recurrences in patients with Stage I seminoma were more common in those managed by surveillance (23 of 94) compared with patients receiving adjuvant radiotherapy (9 of 276). Late recurrences (≥ 2 years) occurred only in the surveillance group. The median time to recurrence was 16.2 months (7.2–56 months) for patients managed by surveillance and 18 months (4–21 months) for patients managed by radiotherapy.
Of the 28 recurrences, which occurred in 135 men treated for Stage II seminoma, 7 occurred more than 2 years after diagnosis. Of the seven recurrences, five occurred in patients treated with single-agent carboplatin as part of initial management (three were treated with carboplatin only). The latest recurrence was seen 6 years after initial diagnosis in a patient who was treated initially with single-agent carboplatin and paraaortic/pelvic radiotherapy. In 60 patients treated for Stage III–IV seminoma, all 13 recurrences were identified within 2 years of initial diagnosis.
Recurrence Pattern in Nonseminoma Patients
In patients with Stage I NSGCTs, late recurrence is uncommon. After a median follow-up of 8 years, of 84 recurrences in 373 patients, 12 occurred after 2 years and only 1 occurred after 5 years (6.7 years after diagnosis).
In contrast, in patients with metastatic NSGCTs (Stage II–IV), a small and continuing risk of very late recurrence was noted. Of 387 patients at risk at 2 years, 26 (6.7%) have had recurrence with 12 (3.1%) men developing recurrence at 5 years or more from initial diagnosis. This approximates to an annual recurrence rate of 1% between 5 and 10 years, with latest observed recurrences occurring at 13.8 and 14 years posttreatment in patients with Stages II and III/IV, respectively.
Features of Very Late Recurrence (after 5 Years)
Overall, 14 patients had recurrence 5 years or more after initial diagnosis. Five hundred eighty-six patients remained on follow-up 10 years after initial presentation, so the maximum recurrence risk is 2.4% (14 of 586). Their characteristics at presentation and recurrence are presented in Tables 6 and 7. There were 12 patients with metastatic NSGCT, one patient with Stage I NSGCT, and 1 patient with metastatic seminoma. Six patients had postchemotherapy resection of residual tumor(s). In six NSGCT patients who had postchemotherapy resection of tumor masses, the resection mass contained differentiated teratoma. In four other patients, the primary tumor contained mature teratomatous elements.
|Patient||Stage||Pathology||AFP||β-HCG||Treatment||Size of residual tumor postchemotherapy (cm)||Postchemotherapy surgery||Postchemotherapy histology|
|1||II||MTU||52||22||PVB + RT||1.5|
|11||IV||MTT||1700||4890||BOP||Not stated||Mediastinal lymph node dissection||MTD|
|14||II||Seminoma||4||< 2||Carboplatin + RT||—|
|RES (ys)||Site of recurrence||Maximum diameter of tumor at recurrence (cm)||Routine follow-up detection||Method of detection||Recurrence histology||AFP at recurrence||β-HCG at recurrence||Treatment at recurrence||Follow-up (mos)||Outcome||Current status|
|1||13.6||Paraaortic||3||Yes||Marker||Yolk sac tumor||27||Normal||Surgery||61||Complete response||NED|
|2||12.6||Paraaortic||7||Yes||Marker, symptoms||—||134,460||Normal||Chemotherapy at other center||34||Further relapse||Lost to FU|
|3||10.2||Paraaortic, bone, brain, lung||4||No||Symptoms||—||85||Normal||High-dose chemotherapy||4||Toxic death||Deceased|
|4||5.6||Lung, neck||2||Yes||Chest X-ray, physical examination||TD||Normal||Normal||Surgery||134||Complete response||NED|
|5||13.9||Paraaortic||2||No||Symptoms||MTU||4450||Normal||Surgery IPE||29||Two further recurrences||Deceased|
|6||9.9||Buttock||10||No||Symptoms||Carcinoma?||Normal||Normal||BOP IPE RT||16||Progressive disease||Deceased|
|7||11.2||Mediastinum||3||Yes||Marker, chest X-ray||MTU||57||Normal||Surgery||42||Complete response||NED|
|8||6.7||Paraaortic, neck, mediastinum||4||Yes||Physical examination||MTU||Normal||Normal||CEB × 3 C-BOP IPE × 2 surgery||93||Complete response||NED|
|9||5.6||Paraaortic||3.5||Yes||Marker||MTU||52||Normal||BOP surgery BEP RT||39||Further recurrence||Deceased|
|10||6.8||Retrocrural, lung||1||Yes||Marker||—||113||Normal||BOP IPE surgery||90||Complete response||NED|
|11||7.7||Mediastinum, paraaortic||2||Yes||Marker||TD||32||Normal||IPE × 2 surgery||56||Complete response||NED|
|12||6.2||Neck||5||Yes||Physical examination||TD||Normal||Normal||Surgery||65||Complete response||NED|
|13||7.5||Retrocrural, paraaortic||2.5||No||symptoms||MTI adenocarcinoma||Normal||Normal||Surgery × 2||34||Complete response||NED|
|14||5.9||Supraclavicular; mediastinum||3||No||Symptoms||Seminoma||Normal||Normal||BEP × 4||34||Complete response||NED|
In two patients with residual tumors larger than 2 cm, surgery was not performed. One patient with a residual paraaortic mass of 3 cm declined surgery. His recurrence developed more than 10 years later in the paraaortic region, as well as in the bone, brain, and lung. The other patient had numerous residual lung nodules following chemotherapy, which were unresectable. His recurrence 11.2 years later was confined to the mediastinum.
Seven patients had elevated α-fetoprotein (AFP) levels at recurrence. None of the patients with recurrence had elevated β-human chorionic gonadotrophin (β-HCG) levels. Lymph node recurrence alone was seen in 10 of 14 patients. In all but one patient, who had recurrence with a carcinoma on a background of teratoma, lymph node disease was a component of recurrence. The paraaortic region was the most common site of recurrence in 8 of 14 men. The average size of the largest tumor at recurrence was 3.5 cm (1–10 cm).
In at least five patients (36%), recurrence was detected at an unscheduled visit arranged by patient due to symptoms. This was the case in the single patient with the late recurring seminoma.
Of 14 patients, 10 had a surgical resection as part of their management at recurrence. Five patients were treated with surgical resection only and all of them were disease free at the time of analysis. Five other patients were treated with chemotherapy and surgery, three of them being disease free at the time of analysis. Treatment at very late recurrence rendered 9 of 14 patients disease free, 8 having surgery as part of their treatment. The patient with the late recurring seminoma achieved a complete response with chemotherapy only. One patient died of treatment-related toxicity, three died after disease progression, and one was lost to follow-up.
The purpose of follow-up after successful treatment of testicular germ cell tumors is to detect recurrence and to monitor and treat complications. The method of follow-up in terms of frequency, duration, and technique has been the subject of debate.1, 5 The main objective of our study was to determine the natural history of recurrence of testicular germ cell tumors and to critically review our current follow-up schedule.
Time to Recurrence
Recurrences are uncommon after 2 years and are even less common after 5 years. Our findings according to histologic and stage subtypes are addressed below and compared with reports in the literature.
In patients with Stage I disease treated with adjuvant radiation, the latest recurrence in our study was seen at 21 months. A similar picture is seen in other series with recurrences after 2 years being extremely rare.6, 7 However, very late recurrences after 5 years have been reported.8 The possibility of a new extragonadal primary tumor rather than recurrence after a primary tumor especially in mediastinal recurrences has been raised.8
In patients with Stage I disease who are managed by surveillance, most recurrences happen in the first 2 years. However, recurrences do occur between 2 and 5 years (the latest recurrence developed at 56 months in our series). In two large series reported by Warde et al.9 (148 patients) and von der Maase et al.10 (261 patients), the median time to recurrence was 15 and 14 months and the latest recurrences occurred at 37 and 61 months, respectively.
A similar picture is seen for Stage II patients with one recurrence in 93 patients at risk of recurrence at 5 years. This patient was treated with a single cycle of carboplatin and dogleg radiotherapy.11 Warde at al.9 reported similar results in this patient group with no recurrences developing after 38 months in 99 patients.12 However, there are anecdotal reports in the literature of recurrences after 5 years.13 As the denominator is usually unknown, the risk is difficult to determine. The diagnostic difficulty of distinguishing a new primary extragonadal germ cell tumor also exists.8, 13
Following chemotherapy in patients with more advanced disease (Stage III and IV), our results suggest that the risk of recurrence is very low after 2 years. However, the smaller number of patients in this subgroup means that these conclusions must be considered tentative. Unfortunately, there are little data in the literature about this group.
Our results, which suggest that recurrence after 24 months is uncommon in patients with Stage I NSGCTs, are supported by Sogani et al.14 In their review of 963 patients, they identified only five recurrences after 2 years. Despite this, very late recurrence even after 10 years has been reported.2, 15 The contribution of routine follow-up and the role of different methods for detection of recurrence are uncertain.
Our data suggest that after chemotherapy in patients with metastatic NSGCTs, a small, but continuing risk of recurrence persists. This is supported by current literature, which includes a number of case reports of late recurrences. Two series confirm these findings. Baniel et al.1 reported on 47 patients with testicular germ cell tumors with recurrences after 5 years. The majority had NSGCTs and clinical and/or pathologic evidence of metastatic disease. An actuarial denominator to calculate the risk of recurrence at different time points is missing, as most patients were referred after recurrence. A similar report by Gerl et al.3 also confirmed the small but continuing risk of late recurrence in patients with metastatic (Stage II–IV) NSGCTs.
It would be useful to identify the patients who are at a particular risk of recurrence. Our univariate analysis of recurrences after 2 years identified advanced stage and positive tumor markers at presentation as risk factors, although only the presence of teratomas in surgical specimens after chemotherapy is significant in multivariate analysis. The small number of events precludes an analysis of recurrences after 5 years.
It is widely believed that late recurrence in patients with NSGCTs is due to residual differentiated teratoma elements after initial chemotherapy treatment. Patients who had surgery after initial chemotherapy and developed recurrence after 5 years all had teratoma differentiated in the original surgical specimen. We believe that all patients with metastatic NSGCTs are at risk of late recurrence and should be followed up for life.
Detecting Late Recurrence
Although a small risk of late recurrence exists in subsets of patients with successfully treated testicular germ cell tumors, the value of an annual or biennial follow up schedule is debatable. If the interval between clinic visits is too long, patients may develop symptoms between visits and recurrence is detected at unscheduled visits. In our series, 5 of 15 patients with recurrence after 5 years developed symptoms leading to an unscheduled clinic visit and subsequent detection of recurrence. In a report by Borge et al.2 13 of 15 recurrences after 3 years were detected at these nonroutine clinic visits.
Role of serum tumor markers in detecting late recurrence
Seven (53%) of our patients with recurrences after 5 years had elevated markers, which was the primary method of detection in six patients. Only raised AFP levels were observed in our series. This is consistent with other series, which suggest that β-HCG production is rare in patients with late recurring tumours.3 The reason for this phenomenon is unknown. It could be related to the potential origin of late recurring differentiated teratomas. Alternatively, it could be that β-HCG–producing cells grow faster and are more chemosensitive, with a tendency of earlier recurrence. This is supported by Gerl et al.3 who observed that AFP-producing tumors are relatively slow growing. Median AFP doubling times in a group of patients with increasing AFP levels on surveillance following primary chemotherapy was 155 days.3 The implication of this observation is that HCG may be of limited value in patients on long-term follow-up.
Role of chest X-ray
Chest X-rays are performed routinely in our long-term follow-up. In two of our patients, chest X-ray was part of the initial detection method. However, one patient also had an elevated marker. In another patient, an abnormal finding on chest X-ray led to thoracotomy, but only necrosis/fibrosis was found on histologic examination. The contribution of routine chest X-ray to patient management in long-term follow-up is minimal.
A role for cross-sectional imaging
In this analysis, the most common site of recurrence was the paraaortic region. Our routine follow-up does not include cross-sectional abdominal imaging. Ravi et al.5 suggested interval imaging with abdominal ultrasound at 5, 10, and 20 years posttreatment. This is an interesting approach. However, ultrasound is inferior to computed tomography (CT) scans for detecting minimally enlarged lymph nodes.16 CT scans would be a more appropriate choice and could be used as an exit examination if discharge from routine follow-up is contemplated. In our series, based on tumor size at the time of recurrence detection, CT scanning would have been able to identify almost all patients with very late recurrence. Follow-up based on CT scanning will have resource implications.
Treatment and Outcome of Late Recurrence
It has been postulated that late recurring disease (arbitrarily defined as happening after 2 years) has a different natural history when compared with untreated primary tumors. It is characterized by slow growth, production of AFP, chemoresistance, and poor prognosis.1, 3, 15 Long-term disease-free survival rates of 25.9% (Baniel et al.1: 81 patients) and 36% (Gerl et al.3: 25 patients) have been reported for patients with recurrence after 2 years. In contrast, 69% of our patients remain disease free at the time of analysis. One reason for this discrepancy may be that these series included patients who received initial treatment at other centers (20 of 81 in Baniel's series and 7 of 25 in Gerl's series). These patients might represent a particularly poor prognostic subset referred to a specialist center for further treatment. Our series is a genuine single-institutional experience and probably a better representation of the natural history of these tumors. A reflection of this is the predominance of surgery, either alone (five patients) or in combination with chemotherapy (five patients), utilized in our series. This suggests a predominance of patients with disease localized to the retroperitoneum in this consecutive series. The good prognosis of patients managed by surgery alone suggests that surgical resection of localized disease is appropriate management for localized recurrences. The value of adjuvant chemotherapy in this situation is uncertain. With the exception of the late seminoma recurrence, no patients were cured without the addition of surgery.
It is worth mentioning that a center's policy on paraaortic lymph node dissection after chemotherapy may potentially affect the incidence and the histology of late recurrence. A more radical approach (performing more extensive resections and utilizing a smaller residual tumor size cutoff) may lead to a lower incidence of late recurrence. With a more conservative approach to lymph node dissection, an increased incidence of late recurrence, a higher prevalence of differentiated teratomas at late recurrence, and a better outcome with treatment would be expected.
Implications for Follow-Up
It is difficult to determine a specific cutoff level for the risk of late recurrence after which long-term follow-up becomes unnecessary. Other issues, such as long-term effects of treatment and psychological aspects of follow-up on patients, need to be addressed. Our study is unique compared with previously reported series. All patients had their initial management at one institution and the actuarial recurrence-free survival rates are more reliable. In patients with seminomatous or Stage I NSGCTs at presentation, the risk of recurrence after 5 years is extremely low and the yield of routine follow-up is very small. However, based on these data, prolonged follow-up for patients presenting with metastatic NSGCTs is appropriate.
On the basis of our data and the published literature, we believe that routine follow-up after 5 years is not required for patients with seminoma at any stage and patients with Stage I NSGCTs. For patients with Stage II–IV NSGCTs at presentation, there is a continuing low risk of recurrence and life-long follow-up should be considered. We will continue to report on this subgroup of patients.
- 4Investigations and staging: general aspects and staging classification. In: PeckhamMJ. The management of testicular tumours. London: Edward Arnold, 1971: 89–101..