The results of this study were presented in part at the 45th Annual Meeting of the American Society of Clinical Oncology, June 7, 2010; Orlando, FL.
Refractory germ cell tumor (GCT) patients have a poor prognosis and limited treatment options. The identification of novel active agents may be impaired by use of response as the primary endpoint in phase 2 trials. Improved endpoints could enhance the development of new effective agents.
The characteristics and outcome of refractory GCT patients enrolled in 7 single-agent phase 2 trials conducted at Memorial Sloan-Kettering Cancer Center from 1990 to 2008 were reviewed. The study agents were suramin, all-transretinoic acid, topotecan, pyrazoloacridine, temozolomide, ixabepilone, and sunitinib. The major endpoints evaluated were response, progression-free survival (PFS), and overall survival (OS).
Ninety patients (87 male, 3 female) were treated. The primary tumor site was testis in 65 patients, mediastinum in 17 patients, retroperitoneum in 4 patients, and other in 4 patients. Eighty-six patients had nonseminoma, and 4 patients had pure seminoma. Best responses were 1 (1%) partial response (ixabepilone), 15 (17%) stable disease, and 74 (82%) progressive disease. Median PFS and OS were 1.0 month (95% confidence interval [CI], 0.8-1.3) and 4.7 months (95% CI, 3.5-6.4), respectively. Eighty-six of the 90 patients have died. The 12- and 16-week PFS rates were 9% (95% CI, 3-15%) and 6% (95% CI, 1%-11%), respectively.
Germ cell tumors (GCTs) are considered a model for the treatment of other malignancies because of the high proportion of patients who can achieve cure even in the face of widely metastatic disease. Currently, more than 70% of such patients are cured with initial cisplatin-based chemotherapy with or without adjunctive surgery.1 Furthermore, 30%-50% of the remainder can still achieve durable remissions with salvage approaches, including high-dose chemotherapy.1 In contrast, nearly all patients progressing after high-dose chemotherapy (referred to as relapsed/refractory), will ultimately die from progressive disease. An average of more than 35 years of life is lost when a patient dies from GCT, well over a decade longer than any other adult malignancy.2
Current therapeutic options for patients with relapsed/refractory GCT are inadequate. Complete responses range from 0% to 10% with salvage regimens such as gemcitabine plus oxaliplatin3-5 or oral etoposide,6 and these responses are rarely durable. In addition, salvage surgery (sometimes referred to as desperation surgery) is only effective for a small percentage of the rare group of patients with a solitary site of residual disease.
These realities underscore the need to identify new active agents against GCT, which has been a major focus of research at Memorial Sloan-Kettering Cancer Center over the past 2 decades. Whereas historic oncologic drug development has focused on cytotoxics, most agents in the current pipeline are considered targeted therapies, many with cytostatic properties. Such agents may produce clinical benefit through induction of prolonged periods of disease stabilization rather than objective responses. Objective response, the traditional primary endpoint for phase 2 trials in GCT, may therefore be suboptimal for assessing these agents, and improved endpoints for assessment of clinical benefit are therefore needed.7 With this in mind, we sought to describe the clinical characteristics and outcome of patients with relapsed/refractory GCT treated at our institution for the purpose of defining progression-free survival (PFS) and overall survival (OS) as new endpoints for phase 2 and 3 trial design.
MATERIALS AND METHODS
From 1990 to 2008, 7 phase 2 trials8-14 with novel single agents were conducted at Memorial Sloan-Kettering Cancer Center for patients with relapsed/refractory GCT. We retrospectively reviewed the characteristics and outcome of refractory GCT patients enrolled in these 7 trials. The study agents (with year of study initiation) were suramin (1990), all-transretinoic acid (1991), topotecan (1993), pyrazoloacridine (1997), temozolomide (2000), ixabepilone (2003), and sunitinib (2007). In general, clinical evaluation of these agents was supported by prior demonstration of efficacy in vitro or in phase 1 studies.8-14 Eligibility for these protocols varied slightly, but in general, patients were included only if they were deemed incurable by standard treatments at the time of the study. Beginning in 1993, all patients were also required to have already received, refused, or be deemed ineligible for high-dose chemotherapy. Ineligibility criteria for high-dose chemotherapy included significant comorbidities, poor renal function, inability to mobilize stem cells, teratoma with malignant transformation, and poor performance status.
Clinical characteristics were obtained based on a query of the data recorded and entered into our institution's central research database at the time the trial was conducted. Data were cross-checked for accuracy against those previously reported at the time the trial was published and were supplemented by review of each individual patient's electronic medical record. This review included outside correspondence to ensure that therapies patients had received at outside institutions were not missed. For patients treated on more than 1 study, baseline characteristics and outcome to the first trial therapy were used. Survival data was based on review of the electronic medical record and central research database, which incorporate data from the Cancer Registry and the Social Security Death Index (updated weekly).
The major endpoints evaluated were response, PFS, and OS. Responses were divided into complete response, partial response, stable disease, and progression of disease. The criteria used to categorize response types varied between the studies as a result of changes over time. World Health Organization criteria were used initially but were later replaced by the Response Evaluation Criteria in Solid Tumors. In addition, the frequency of imaging assessments varied, but was at a minimum between every 4 and 8 weeks. If clinical progression (based on tumor marker rise and deteriorating status/symptoms) occurred before the planned imaging assessments, the date the patient was taken off the study was used as the official date of progression. Patients who were removed from the study due to toxicity were censored but were recorded as nonprogressors.
PFS was defined as the time from treatment initiation until disease progression or death, whichever occurred first. OS was defined as the time from treatment initiation until death from any cause. PFS and OS curves, as well as 95% confidence intervals (CIs), were estimated according to the Kaplan-Meier method. A log-rank test was used to determine whether PFS and OS curves differed among groups defined by each demographic and clinical variable. Twelve-week PFS estimates and 6-month OS estimates with CIs were estimated using the log-log method were generated.
Between 1990 and 2008, 90 patients were treated in the 7 studies (Table 1), including 4 patients who were enrolled in more than 1 study. The majority of patients were male (n = 87) and Caucasian (n = 85), but 3 women and 5 non-Caucasian patients were also included. The median age was 35 (range, 17-64), and the primary tumor site was testis in 65 patients, mediastinum in 17 patients, retroperitoneum in 4 patients, and other in 4 patients. Eighty-six patients had nonseminoma, and 4 had pure seminoma. Four patients had marker-only disease, and the remaining 86 patients had radiographic evidence of disease with or without serum tumor marker elevation. The median number of metastatic sites was 2 (range, 2-5), and the most commonly involved areas were the lung, retroperitoneal lymph nodes, liver, and other (non-retroperitoneal) lymph nodes.
Table 1. Phase 2 Studies in Patients With Relapsed/Refractory Germ Cell Tumor from 1990 to 2008
The total number of patients adds up to 94 because 4 patients were treated on more than 1 study.
All patients had been treated previously with cisplatin-based chemotherapy. The median number of prior lines of chemotherapy was 3 (range, 1-6). Fifty-six (62%) patients were confirmed to have progressed despite prior treatment with high-dose chemotherapy. Forty-six (51%) patients had received paclitaxel as part of 1 or more of their prior treatment regimens. Many patients who were not previously treated with paclitaxel were treated in the early 1990s before the identification of the activity of this agent against GCT. A full set of patient characteristics is provided in Table 2.
Table 2. Patient Characteristics
AFP, alpha-fetoprotein; HCG, human chorionic gonadotropin; LDH, lactate dehydrogenase.
Data are expressed as no. (%) unless specified otherwise.
Only 1 (1%) of the 90 patients achieved an objective response, which was a partial response to ixabepilone. Fifteen (17%) patients achieved a best response of stable disease, including 3 patients each on sunitinib, ixabepilone, topotecan, and temozolomide; 2 patients on all-transretinoic acid; and 1 patient on suramin. The overwhelming majority of patients (n = 74, 82%) had progressive disease as their best response, an indication of disease progression prior to or at the first response assessment. This included all 13 patients treated with pyrazolacridine (Table 3).
Table 4. Treatment and Outcome for Patients With Progression-Free Survival >12 Weeks
No. of Patients
4.5 mo, 6.8 mo
3.7 mo, 3.7 mo
Of note, the 1 responding patient had a primary mediastinal seminoma and had received less extensive pretreatment than the majority of patients included in the 7 studies. His prior treatment included 4 cycles of bleomycin plus etoposide plus cisplatin followed by mediastinal mass resection and then 2 additional cycles of adjuvant EP for residual viable seminoma found at the time of surgery. He had not received prior high-dose chemotherapy or conventional-dose treatment with either paclitaxel or ifosfamide, because these treatments had been offered but declined by the patient due to toxicity concerns.
PFS and OS
Two patients were taken off the study for toxicity without documented progression, including 1 patient treated with suramin and 1 patient treated with ixabepilone. These patients were censored at the time of study discontinuation. The remaining 88 patients all progressed during treatment. The median PFS for the entire cohort was 1.0 months (95% CI, 0.8-1.3). At 12 weeks, <9% (95% CI, 3%-15%) of patients were alive and progression-free, and by 16 weeks, this figure dropped below 6% (95% CI, 1%-11%) (Table 3 and Figure 1). The number of patients who remained progression-free at 12 weeks are listed in Table 4 according to the investigational agent they received.
Overall, 86 patients have died, 2 have been lost to follow-up, and 2 are alive and disease-free. Of these latter 2 patients, 1 was rendered continuously disease-free by salvage resection of a lung nodule after progression on sunitinib, now 33+ months from initiation of sunitinib. The other patient progressed on topotecan before receiving paclitaxel as a single agent and achieving a complete response to 6 cycles. This patient remains continuously disease-free since treatment with paclitaxel, now 190+ months from initiation of topotecan. The median OS was 4.7 months (95% CI, 3.5-6.4) and is illustrated in Figure 2.
Clinical factors (Table 1) were analyzed for an association with an adverse outcome. Elevated lactic acid dehydrogenase (LDH) level was the only variable that predicted statistically significant prolonged OS; 25% (95% CI, 13-38) of patients with elevated LDH levels achieved 6-month OS versus 57% (95% CI, 41-70) of those with normal LDH levels. For PFS, an elevated LDH level (12-week PFS, 4% vs 13%; P = .012) and a human chorionic gonadotropin level ≥1000 mIU/mL (12-week PFS, 0% vs 10%; P = .026) both correlated with adverse outcome, whereas the presence of unusual sites of metastases (spleen, small intestine, esophagus, skin, heart) was associated with a higher likelihood of achieving 12-week PFS (24% vs 5%; P = .022). A trend toward inferior 12-week PFS (P = .06) and 6-month OS (P = .07) was also observed for patients with Karnofsky performance status of 70% versus ≥80%. Elevated alpha-fetoprotein level and prior treatments received were not prognostic for OS or PFS.
Current treatment options for patients with relapsed/refractory GCT include palliative single agents and combinations, salvage surgical resection, and palliative radiation therapy to osseous or central nervous system metastases. However, these options are inadequate, with nearly all patients destined to die from progressive disease. Furthermore, relapsed/refractory GCT patients are typically young adult men in the prime of their lives, and their deaths are responsible for the greatest average number of life years lost of any malignancy.2
Therefore, there is a need to identify and develop new active agents against GCT. However, there are several barriers to achieving this goal, including the long time to complete studies due to the small number of patients seen at any individual site and a historic lack of international cooperation in studying the treatment of this patient population. In addition, with the increasing development of targeted therapies with cytostatic rather than cytotoxic properties, it is possible that traditional clinical trial endpoints such as objective response are suboptimal.7 Although cytostatic targeted agents could provide significant clinical benefit, including prolongation of survival through induction of durable periods of disease stabilization, they would be deemed ineffective if objective responses were used to assess activity. Therefore, novel endpoints are needed for a phase 2 trial design in this population to prevent premature rejection of such agents. We hypothesized that PFS and OS represent suitable endpoints for the study of novel agents in patients with relapsed/refractory GCT.
The incorporation of new endpoints into phase 2 trials depends on reliable historical data. For example, in Simon's 2-stage design15 for single-arm phase 2 trials, both uninteresting (null hypothesis, P0) and interesting (alternative hypothesis, P1) levels of activity must be chosen. The specific rates selected for the null and alternative hypotheses are a critical part of trial design, directly affecting whether a drug is deemed promising for further study as well as the sample size and duration of the trial. Thus, to devise a trial for patients with relapsed/refractory GCT with PFS or OS as the primary endpoint, historical data on PFS and OS in a similar patient population is necessary, in particular when defining the null hypothesis.
In this study, we reviewed the characteristics and outcomes of 90 patients with relapsed/refractory GCT treated on 7 negative phase 2 trials at Memorial Sloan-Kettering Cancer Center between 1990 and 2008. All 7 agents were declared ineffective; therefore, this population represents a good historical control group for which future studies can be based. Our primary observation is that patients with relapsed/refractory GCT are very unlikely to achieve prolonged PFS (>12 weeks) in the absence of treatment with an effective therapy. This is in contrast to other malignancies such as renal cell carcinoma or melanoma, where prolonged periods of stable disease can be observed in some patients not receiving any form of active therapy. We identified a median PFS of only 1 month in this population and demonstrated that <9% and <6% of patients remain progression-free after 12 and 16 weeks, respectively, with an overall median survival of only 4.7 months. We also identified elevated LDH level as the sole factor prognostic for both decreased PFS and OS in this patient population.
These data are well-suited for application to the design of phase 2 trials in patients with relapsed/refractory GCT. For example, based on our data, a Simon's 2-stage design could use 12-week PFS as its primary endpoint with a cutoff value of 10% as the uninteresting outcome (P0) and 25% as warranting further study (P1). With an alpha of 0.05 and a beta of 0.20, a trial according to Simon's optimum 2-stage design15 would accrue 18 patients in the first stage with a stopping rule to close the study early unless 2 or more patients remained progression-free at 12 weeks.. If ≥3 of the initial 18 patients achieved at least 12-week PFS, the trial would expand to a total of 43 patients and the agent deemed worthy of further study if ≥8/43 (19%) patients were alive and progression-free at 12 weeks.15
It might be argued that the low rate of 12-week PFS proposed as the historical comparator for future phase 2 trials might lead to false acceptance of an inactive drug for larger study based on chance enrollment of a favorable population. However, the universal poor outcomes of the 90 patients described herein—despite heterogeneous clinical factors and treatment with multiple different agents over nearly 20 years—suggest that this is unlikely. Furthermore, an analysis of various clinical factors for association with a more favorable or adverse outcome among this group revealed a very limited potential to discriminate patients. As mentioned above, only elevated LDH was prognostic for both PFS and OS. This is in agreement with another retrospective analysis of 54 GCT patients who underwent systemic treatment after progression to high-dose chemotherapy in which no clinical factor could be identified as prognostic for survival.16
Simon's 2-stage design is particularly advantageous to trials conducted in rare disease states such as relapsed/refractory GCT, where patient accrual can be challenging. The early stopping rule for inefficacy allows investigators to quickly close a trial of an inactive agent and move on to a new study that may hold more promise. More widespread adoption of this strategy could therefore improve the efficiency of trial conduct in this setting and expedite the development of novel active agents for GCT. Similarly, the early stopping rule limits the number of patients exposed to agents unlikely to offer therapeutic benefit, allowing patients to pursue more worthwhile treatments, and potentially enhancing accrual to other more promising trials.
In conclusion, these data illustrate the rapid progression that nearly all patients with relapsed/refractory GCT experience when treated with ineffective drugs and indicate that when prolonged stable disease (≥12 weeks) is achieved, it is likely to be a manifestation of effective drug therapy. Therefore, we recommend that 12-week PFS (with comparison with the 9% benchmark rate reported herein) be used as the primary endpoint for future phase 2 clinical trial design in this population. This is particularly important for the study of agents with cytostatic antitumor properties to avoid the inappropriate dismissal of potentially beneficial novel therapeutics. Similarly, these data support using OS as the primary endpoint for phase 3 trials in this population, with a median survival of approximately 5 months expected for the control arm.
Supported by the Sidney Kimmel Center for Prostate and Urologic Cancers (New York, NY) and the Craig Tifford Foundation.