First salvage treatment of germ cell tumor patients with bone metastases: retrospective analysis of a large international database

Abstract

Purpose

To retrospectively analyze clinical characteristics, prognostic factors, and optimal treatment of patients with bone metastases (BM) of germ cell tumors (GCT) at first relapse.

Methods

One hundred and four GCT patients with BM were identified from the IPFSG database containing 1,594 patients at first relapse. Within this database, all patients experienced unequivocal relapse/progression after cisplatin-based chemotherapy and had received either conventional (CD-CTX) or high-dose chemotherapy (HD-CTX) as first salvage treatment.

Results

At relapse, eight patients (8 %) had BM only, concomitant relapse with lung, brain, liver and/or nodal metastases were present in 40 (39 %), 6 (6 %), 27 (26 %), and 69 (66 %) pts, respectively. Patients clustered over all IPFSG subgroups, and the IPFSG score could be confirmed. Salvage treatment was CD-CTX in 35 and HD-CTX in 69 patients. Overall response (CR, PR) rate to salvage chemotherapy was 81 % (HD-CTX) versus 43 % (CD-CTX; p < 0.001). Median follow-up was 14 months (mos; range 1–161). Both, median PFS and OS, were higher after HD-CTX compared to CD-CTX [PFS 9 (95 % CI 6–12) vs. 5 (3–7) mos (p < 0.01); OS 18 (12–24) vs. 13 (8–18) mos (p = 0.078)].

Conclusions

GCT patients relapsing with BM have a dismal outcome. Retrospectively, first salvage HD-CTX seems to improve treatment response and outcome. Further evaluation of characteristics and treatment of GCT patients with BM is warranted.

Introduction

Bone metastases (BM) are a rare site of metastatic disease in germ cell tumor (GCT) patients, occurring in approximately 3–9 % at primary GCT diagnosis. Mostly, BM occur as part of widespread systemic disease, involving the lungs, retroperitoneal lymph nodes and/or the liver, or in patients with primary mediastinal GCT (Johnson et al. 1976; Hitchins et al. 1988; IGCCCG 1997; Einhorn 2002; Oechsle et al. 2012). BM are often localized in the midline of the body, i.e., spine, pelvis, and ribs, and less frequently at other bone localizations. The presence of BM classifies nonseminoma patients as ‘poor prognosis’, and seminoma patients as ‘intermediate prognosis’ according to the International Germ Cell Cancer Cooperative Group (IGCCCG) criteria (IGCCCG 1997; Jamal-Hanjani et al. 2013). Overall, prognosis of patients progressing during or relapsing after cisplatin-based first-line therapy is substantially worse, but survival differs markedly (Kollmannsberger et al. 2008; Koychev et al. 2011). Risk factors to predict outcome have recently been identified by the International Prognostic Factors Study Group (IPFSG) (Lorch et al. 2010). Optimal management at relapse has not been determined to date, but different regimens of combination chemotherapy (CTX), i.e., TIP, VeIP, VIP, or GIP have shown efficacy and provide a curative option for first salvage treatment (Einhorn 2012; Oldenburg et al. 2013; Fizazi et al. 2014). High-dose CTX seemed to beneficially impact outcomes in numerous retrospective analyses, but until now this benefit was not confirmed by a randomized, prospective phase III clinical trial (Oldenburg et al. 2013).

In particular, data on characteristics, risk factors, optimal treatment approach, and outcome of GCT patients with BM are largely lacking. Only a small number of case reports or small analyses with at maximum 40 GCT patients with BM have been published to date (Asthana et al. 1988; Hitchins et al. 1988; Bosco et al. 1994; Arnold et al. 2000; Watanabe et al. 2004; Miyake et al. 2005; Uygun et al. 2006; Benedetti et al. 2006; Aldejmah et al. 2007; Ozan et al. 2009; Nachankar et al. 2013; Wudhikarn et al. 2013; Jamal-Hanjani et al. 2013).

The purpose of this retrospective subgroup analysis of 104 patients identified from a large international database of a total of 1,594 male GCT patients with relapse after first-line CTX was to determine clinical characteristics and treatment outcome. Furthermore, we retrospectively analyzed the impact of different salvage approaches comprising both conventional dose CTX (CD-CTX) and high-dose CTX (HD-CTX), as well as the role of additional local treatment approaches, i.e., radiotherapy and/or surgery, on the outcome of this subgroup.

Patients and methods

Study population

104 GCT patients with BM at first relapse were identified from the International Prognostic Factor Study Group (IPFSG) database. This preexisting database contains data of 1,594 patients with unequivocal relapse or progression after at least three cycles of cisplatin-based first-line CTX from 38 international centers. All patients had received cisplatin-based CD-CTX or carboplatin-based HD-CTX as salvage treatment. A detailed description of data acquisition and handling has previously been published (Lorch et al. 2010).

Inclusion criteria

Inclusion criteria were the following: male sex, age ≥15 years; metastatic GCT defined either histologically and/or by unequivocal serum tumor markers [beta human chorionic gonadotropin (ß-hCG) or α-fetoprotein (AFP)]; first-line CTX after January 1st, 1990; three or more cycles of cisplatin-plus etoposide-based first-line CTX; no primary HD-CTX; unequivocal relapse or progression after first-line CTX; no previous salvage CTX; first salvage treatment with either CD-CTX or carboplatin-based HD-CTX; and availability of baseline and follow-up information to sufficiently calculate primary and secondary outcome variables. To identify our BM subgroup, patients, who either had solely BM or concomitantly relapsed at other locations, were chosen. Prior radiotherapy and/or surgery as part of first-line treatment were no exclusion criteria.

Statistical analysis

Median progression-free survival (PFS) was the primary study end point, and PFS was defined from the onset of salvage CTX until progression of disease or last follow-up. Patients dying without progression were censored at the time of death. Median overall survival (OS), 1-year PFS and 1-year OS were secondary study end points. OS was calculated from the onset of salvage CTX until death. Patients lost to follow-up were censored at the time of last visit.

All statistical analyses were conducted using SPSS software version 22 (IBM, USA). Calculation of correlations of characteristics between different subgroups was performed using the χ ² test for categorical variables. Survival rates were analyzed using the method of Kaplan–Meier, and the log-rank test was performed to analyze the significance of differences between the groups.

Results

Patient characteristics at primary diagnosis and primary treatment

A total of 104 of 1,594 patients (7 %) presented with BM at first relapse. Median age of patients at initial GCT diagnosis was 30 years (range 16–53). Primary histology was pure seminoma in 16 patients (15 %), nonseminoma or mixed GCT in 83 patients [80 %; n = 5 unknown (5 %)]. At initial diagnosis, 18 (17 %) and 86 (83 %) of the patients were classified as ‘intermediate’ and ‘poor prognosis’ according to the IGCCCG criteria (IGCCCG 1997). The time point of first occurrence of BM could be extrapolated in 99 patients. Of these, 84 patients (85 %) had progressive BM at relapse and therefore have had primary bone lesions, while 15 patients (15 %) on the contrary have developed BM de novo at first relapse. Tumor marker values at initial diagnosis were not predictive for later development of BM. Overall response rate (ORR) to first-line cisplatin-based CTX was 89 %. Median progression-free survival (PFS) after primary treatment was 9 months (95 % CI 8–10).

Patient characteristics at relapse

Eight patients (8 %) relapsed exclusively with BM prior to salvage treatment, of which all eight patients already had BM at primary GCT diagnosis. The other patients synchronously relapsed in the lungs, liver, brain and/or lymph nodes in 40 (39 %), 27 (26 %), 6 (6 %), and 69 patients (66 %), respectively. Progressive abdominal, mediastinal and/or cervical nodal tumor manifestations were found in 38 (37 %), 22 (21 %), and 9 patients (9 %). A total of 18 patients (17 %) had suffered late relapse (relapse >2 years after onset primary treatment). At the time of relapse, histological confirmation was obtained in 37/104 patients (36 %), only. Histology was pure seminoma in 7 patients, and nonseminoma or mixed GCT in 30 patients, including mature teratoma components in 4 patients and solely mature teratoma in 2 patients. None of the initial seminoma patients relapsed with nonseminoma histology. In total, 12 patients (12 %) had completely normal serum tumor marker values (AFP, ß-HCG, and LDH) at first relapse. Of 24 patients (23 %), at least one value was missing. Elevation of AFP at relapse was found in one seminoma patient and 3/5 patients with unknown primary histology, who were classified as having nonseminoma for the purpose of further analysis.

IPFSG risk stratification

IPFSG risk stratification in our subgroup was as follows: Very low risk in 1 (1 %), low risk in 12 (11 %), intermediate risk in 39 (38 %), high risk in 22 (21 %), and very high risk in 8 patients (8 %). Due to missing values in individual variables, 22 patients (21 %) could not be classified according to IPFSG score. For further details of IPFSG scoring factors, please see Table 1.

Table 1.

IPFSG scoring of the BM subgroup

IPFSG score characteristics	Pts with BM no. (%)	IPFSG score points
Site of primary tumor
Gonadal	91 (88)	0
Extragonadal	9 (8)	1
Mediastinal nonseminoma	4 (4)	3
Prior response to first-line treatment
CR/PRm−	77 (74)	0
PRm +/SD	16 (15)	1
PD	8 (8)	2
Missing	3 (3)	–
Progression-free interval
>3 months	95 (91)	0
≤3 months	9 (9)	1
AFP prior salvage
Normal	48 (46)	0
≤1,000 kU/l	27 (26)	1
>1,000 kU/l	22 (21)	2
Missing	7 (7)	–
ßHcG prior salvage
≤1,000 kU/l	79 (76)	0
>1,000 kU/l	13 (12)	1
Missing	12 (12)	–
Liver, brain and/or bone metastases at relapse (LBB)
Yes	99 (95)	1
Bone metastases only	68 (65)	–
LB without bone metastases at first relapse	31 (30)	–
Missing	5 (5)	–
Primary histology
Seminoma	16 (15)	−1
Nonseminoma or mixed GCT	86 (83)	0
Unknown	2 (2)	–
Final IPFSG prognostic score
Very low risk	1 (1)
Low risk	12 (12)
Intermediate risk	39 (38)
High risk	22 (21)
Very high risk	8 (7)
Incomplete data	22 (21)

CR complete remission, PRm- tumor marker negative partial remission, PRm + tumor marker positive partial remission, SD stable disease, PD progressive disease

Survival analysis with respect to the IPFSG risk categories confirmed the validity of the score, even in the presence of BM as a supposedly very poor risk feature (PFS, p = 0.038; OS, p = 0.046). Only one seminoma patient with BM at initial diagnosis had a very low risk profile, which is why no survival estimates could be calculated for this subgroup. Median PFS was 17 months (95 % CI 13–21) for the low-risk group, 9 months (95 % CI 4–14), 5 months (95 % CI 4–6), and 6 months (95 % CI 4–8) for the intermediate, high, and very high risk groups, respectively. Median OS was 21 months for low-risk patients (range 3–119) and was 27 months (95 % CI 6–48), 14 months (95 % CI 10–18), and 10 months (95 % CI 9–11) for intermediate, high, and very high risk patients, respectively. Kaplan–Meier survival curves are shown in Fig. 1.

Fig. 1.

Kaplan–Meier estimates for PFS (a) and OS (b) according to IPFSG risk categories. Patient numbers in the different risk categories were: Very low risk n = 1, low risk n = 12, intermediate risk n = 39, high risk n = 24, and very high risk n = 8, respectively

Of the 82 evaluable patients, 70 had metastatic spread to liver, brain or bones (LBB) as an independent IPFSG risk factor. 43/70 patients (61 %) only had BM, 23 had accompanying liver, two had brain, and one patient had liver and brain metastases. 1/70 patient had liver metastases only without BM at relapse. To better define the particular impact of BM on survival, we conducted a survival analysis comparing patients with BM only to those with concomitant liver and/or brain metastases. Both PFS and OS did not differ significantly between both subgroups in each IPFSG risk category.

Salvage treatment

Salvage treatment consisted of CD-CTX in 35 (34 %) and HD-CTX in 69 patients (66 %). Of these 69 patients, 31 (45 %) had received single and 38 (55 %) sequential salvage HD-CTX followed by autologous stem cell reinfusion. A total of 12 patients (11 %) underwent additional surgery after salvage CTX; another 5 (5 %) patients underwent both surgery and radiotherapy. Secondary surgery of residuals in the bones was performed only in two patients. Histological examination of these bone lesions revealed vital tumor and mature teratoma in one patient each. Moreover, 29 patients (28 %) received additional irradiation at different localizations, which cannot be further localized, since this detail was not part of the data collection. The impact of additional local treatments, i.e., secondary surgery and/or irradiation of residual tumors, on survival could not be analyzed, as information on irradiated area was not sufficiently documented, and secondary resection of bone lesions was documented in only 2 patients, preventing more detailed analysis. Further information regarding applied salvage treatments is shown in Table 2.

Table 2.

Details of salvage treatment at first relapse

Characteristics	Pts with BM no.	%
Salvage chemotherapy regimens
Conventional dose	35	34
High dose	69	66
Conventional-dose salvage chemotherapy regimens
VIP	8	23
TIP	7	20
VeIP	2	34
Other	8	23
High-dose salvage chemotherapy regimens
CE	26	38
CEC	21	20
CET	6	9
CEI	3	4
Other	13	19
Secondary local consolidation treatment
No further treatment	58	56
Surgery	12	11
Radiotherapy	29	28
Surger + radiotherapy	5	5

VIP etoposide/ifosfamide/cisplatin, VeIP vinblastine/ifosfamide/cisplatin, TIP paclitaxel/ifos-famide/cisplatin, TI paclitaxel/ifosfamide, CE high-dose carboplatin/etoposide, CEI high-dose CE/ifosfamide; CET, high-dose CE/paclitaxel; CEC, high-dose CE/cyclophosphamide

ORR to salvage chemotherapy was 68 % for the entire cohort. Median follow-up was 14 months (range 1–161). Overall, 73 patients (70 %) progressed at a median of 8 months (range 1–161), and 62 patients (60 %) died at a median of 11 months (range 1–95) after onset of salvage treatment. A total of 12 patients (12 %) were lost to follow-up. Kaplan–Meier estimates for entire cohort (n = 104 patients) were a median PFS of 8 months (95 % CI 7–10) and a median OS of 15 months (95 % CI 11–20), respectively. Concordantly, 1-year PFS and 1-year OS were 37 and 59 %, respectively.

Salvage CD-CTX versus HD-CTX

Comparing patients who had undergone CD-CTX versus HD-CTX, ORR was significantly higher after HD-CTX with 81 versus 43 % after CD-CTX (p < 0.001). Moreover, a complete remission was exclusively achieved by HD-CTX and not by CD-CTX (21 vs. 0 %; p < 0.01). Progressive disease despite salvage treatment occurred significantly more often among CD-CTX patients (48 vs. 16 %; p < 0.001). Median PFS was significantly longer after HD-CTX compared to CD-CTX with 9 months (95 % CI 6–12) versus 5 months (95 % CI 3–7; p < 0.01). Median OS did not reach statistical significance with 18 months (95 % CI 12–24) versus 13 months (95 % CI 8–18; p = 0.078). Both, 1-year PFS and 1-year OS were higher in patients receiving HD-CTX compared to CD-CTX, but differences did not differ significantly (1-year PFS 41 vs. 29 %, p = 0.23; 1-year OS 19 vs. 65 %, p = 0.056). Comparing patients who received sequential HD-CTX to those who underwent CD-CTX, both median PFS and OS were significantly improved by sequential HD-CTX. Median PFS was 14 months (range 2–127) after sequential HD-CTX versus 5 months (95 % CI 3–7) after CD-CTX (p < 0.001), and median OS was 21 months (range 3–127) versus 13 months (95 % CI 8–18) for both groups (p = 0.022), respectively. Corresponding Kaplan–Meier curves are shown in Fig. 2. Detailed outcomes after completion of salvage treatment procedures are shown in Table 3.

Fig. 2.

Kaplan–Meier estimates for PFS (A1) and OS (A2) according to intensity of salvage chemotherapy comparing conventional to high-dose salvage CTX (both single and sequential regimens). HD-CTX n = 69 patients, CD-CTX n = 35 patients. Kaplan–Meier estimates for PFS (B1) and OS (B2) according to intensity of salvage chemotherapy comparing conventional to sequential high-dose salvage CTX regimens. Sequential HD-CTX n = 38 patients, CD-CTX n = 35 patients

Table 3.

Treatment outcome after complete salvage treatment

Outcome	CD-CTX no. of pts (%)	HD-CTX no. of pts (%)	p value
ORR	15 (43)	56 (81)	<0.001
CR	0 (0.0)	15 (21)	<0.01
sCR	4 (11)	6 (9)	0.65
PRm-	9 (26)	22 (32)	0.52
PRm+	2 (6)	13 (19)	0.07
SD	3 (9)	2 (3)	0.20
PD	17 (48)	11 (16)	<0.001

ORR overall response rate, CD-CTX conventional dose chemotherapy, HD-CTX high-dose chemotherapy, CR complete remission, sCR secondary CR after local treatment, PRm− tumor marker negative partial remission, PRm+ tumor marker positive partial remission, SD stable disease, PD progressive disease

Single versus sequential salvage HD-CTX

Comparing patients who received either single or sequential HD-CTX, no significant difference in ORR could be detected (ORR 77 % after single vs. 90 % after sequential HD-CTX; p = 0.17). Comparing patients receiving single to sequential salvage HD-CTX, sequential HD-CTX significantly improved PFS compared to single HD-CTX [median PFS 14 (range 2–127) vs. 8 months; (95 % CI 7–9); p < 0.01]. Improvement of OS was of borderline statistical significance [median OS 21 (range 3–127) vs. 15 months (95 % CI 9–21); p = 0.05)]. This translated into a 1-year PFS of 50 versus 29 % (p = 0.078), and a 1-year OS of 71 versus 58 % (p = 0.26) comparing sequential versus single HD-CTX. Corresponding Kaplan–Meier survival curves are shown in Fig. 3.

Fig. 3.

Kaplan–Meier estimates for PFS (a) and OS (b) according to intensity of salvage HD-CTX comparing single to sequential high-dose salvage CTX. Numbers of patients receiving sequential or single HD-CTX were n = 38 versus n = 31, respectively

Comparison of seminoma and nonseminoma patients

Comparing seminoma and nonseminoma patients, seminoma patients showed a trend toward a better outcome. Median PFS of seminoma patients was 17 months (range 3–119), and of nonseminoma or mixed GCT patients 8 months (95 % CI 6–10; p = 0.064), median OS was not reached in seminomas and estimated at 15 months (95 % CI 10–19) for nonseminomas (p = 0.059). This translated into a significantly improved 1-year PFS of 63 % (10/16) versus 33 % (28/86; p = 0.028), and a marginal improvement of 1-year OS of 69 % (11/16) versus 58 % (50/86; p = 0.42) for seminomas. Furthermore, no significant differences in treatment outcome were observed between seminomas and nonseminomas when comparing patients of both subgroups, who had received either CD-CTX or HD-CTX. However, the subgroups are very small, so this result must be interpreted with caution.

Discussion

To our knowledge, this retrospective analysis of 104 patients relapsing with BM after cisplatin-based first-line chemotherapy represents the largest cohort both of GCT patients with BM in general, and particularly in the salvage setting, so far. The strengths of our analysis are the strict eligibility criteria, especially the requirement of contemporary treatment regimens for both first-line and first salvage treatment, the documentation of unequivocal disease progression prior to first salvage treatment to exclude those patients who had undergone salvage treatment as intensification of first-line CTX, as well as the international multicenter approach, which diminishes likely biases of institutional or regional preferences.

Interestingly, the median PFS of 9 months following first-line treatment in our cohort of BM patients was considerably longer compared to the PFS of the entire IPFSG cohort of 1,594 patients, who achieved 3 months, only. This might imply two conclusions, namely an assumably good response of BM to cisplatin-based first-line treatment and/or a comparably slow growth of BM after first-line treatment.

Nevertheless, BM confer a substantially negative prognostic impact on the outcome of first salvage treatment. Both, median PFS and OS after salvage therapy were inferior compared to the entire IPFSG cohort (Lorch et al. 2010). But still, salvage treatment can be successful and long-term survival can be achieved in this particularly poor risk subgroup of patients.

Furthermore, survival differences according to IPFSG risk stratification were confirmed in our BM patient subgroup, which also confirms GCTs with BM as a clinically and prognostically heterogeneous subpopulation, as BM patients clustered over all IPFSG risk groups. Further comparative analysis of PFS and OS between patients with the presence of the LBB risk factor with or without BM did not differ significantly. Thus, BM alone do not seem to confer a significantly pronounced negative impact on survival.

Primary histology seems to impact survival as well, favoring seminoma patients, but differences failed to reach statistical significance due to the small number of seminoma patients in our study population. The clinical characteristics of patients relapsing with BM after cisplatin-based first-line CTX are unspecific, and no predictors for the subsequent development of BM could be identified. However, BM were mostly part of widespread metastatic disease, although eight patients relapsed solely with BM. The fact that all of these eight patients already had BM at initial GCT diagnosis emphasizes the need for a better understanding of how to treat particularly BM, both in the primary and the salvage treatment setting.

Our retrospective analysis seems to suggest that HD-CTX may be superior to CD-CTX for patients with BM at relapse, regarding both PFS and ORR. OS seemed favorable for patients receiving sequential HD-CTX, but this was only of borderline statistical significance. This potential advantage of HD-CTX is in line with the results of the entire IPFSG cohort, in which HD-CTX seemed to be superior to CD-CTX as first salvage treatment for almost all patients, except for those with a low-risk profile (Lorch et al. 2011). On the contrary, the only randomized clinical trial so far by Pico et al. directly comparing CD-CTX to HD-CTX failed to improve both PFS and OS by treatment intensification (Pico et al. 2005). In the only randomized phase III trial so far, which compared single to sequential salvage HD-CTX in a total of 211 patients, sequential HD-CTX seemed to improve survival. However, this did not result from improved therapeutic efficacy, but increased treatment-associated mortality of the single HD-CTX arm using a combination including cyclophosphamide (Einhorn 2012; Lorch et al. 2012). In this analysis, sequential HD-CTX seemed superior to single HD-CTX regimens. We believe that whenever salvage HD-CTX is chosen, two or three cycles of the combination of carboplatin and etoposide should be preferred to a single HD-CTX adding a third agent (Beyer et al. 2013). But, the optimal treatment approach for patients with BM at relapse still remains to be determined, particularly since HD-CTX is not generally accepted as first salvage treatment option.

The optimal utilization of secondary surgery, radiotherapy, or combinations thereof for BM currently remains elusive. Whereas secondary resection of any residual tumor >1 cm in size with normal serum tumor markers in nonseminoma patients is the standard of care (Krege et al. 2008; Oldenburg et al. 2013), secondary resection especially of multiple residual bone lesions is often technically not feasible. In our cohort, particular resection of residual bone lesions was documented in only two cases. Histology revealed mature teratoma in one (who had received salvage CD-CTX), and vital carcinoma in the other patient (who had received salvage HD-CTX), which raises the question whether or not resection of all bony residues is necessary (if technically feasible), or if histological confirmation would be mandatory before the initiation of secondary radiotherapy. The findings do not support the results of a previous retrospective analysis by our working group, in which secondary resected bone lesions after first-line HD-CTX were found to be necrotic in all 4 cases (Oechsle et al. 2012). Nevertheless, in both studies, the numbers of patients were too small to reasonably assess the impact of resection of bone metastases on survival. The database was not constructed to evaluate the impact of different localizations of additional radiotherapy, which is why the localization of radiotherapy was unknown, and no further survival analyses were conducted. In conclusion, the impact of additional surgery and/or radiotherapy of residual bone lesions remains difficult to define, to date.

Finally, some limiting aspects of this analysis have to be considered. The retrospective character of this study limits the conclusive power and only allows descriptive conclusions and generation of hypotheses. The limited median follow-up of 14 months limits the explanatory power of our survival analyses. Furthermore, several potential biases have to be considered. The allocation of patients to CD-CTX and HD-CTX as well as chosen CD-CTX and HD-CTX regimens varied between participating centers and countries, and positive selection favoring HD-CTX may have been confounding patients. The treatment at expert centers may in fact have positively impacted treatment outcomes (Collette et al. 1999). Additionally, in our database, detailed information regarding both localization and number of BM and documentation of symptoms related to BM was lacking. This is why a more comprehensive characterization of the pattern of BM both at initial diagnosis and at relapse of GCT remains elusive. To this end, a large multinational register study of GCT patients with BM has currently been initiated.