Era of discovery (1960 s –1985)

Use of early chemotherapy agents began in the 1960s and often consisted of the serial use of single agents. Measurable clinical activity was demonstrated for bleomycin, vinblastine, actinomycin D and minor activity for anthracyclines and classic alkylators. Simple combination therapy, such as vinblastine and bleomycin, demonstrated occasional complete responses, but few durable remissions.

In 1965, Barnett Rosenberg and colleagues at Michigan State University discovered biologic effects of elemental platinum on Escherichia coli during experiments using a platinum anode to conduct electricity through bacterial cultures. Dr Rosenberg and his team hypothesized that a platinum compound was disrupting the ability of the E coli to divide and went further to identify the platinum species with the most profound effect: cis -diaminodichloroplatinum.

The team at Michigan State further speculated that this compound might merit testing in cancer. They performed in vitro experiments with L1210 leukemia cells as well as animal sarcoma models. Ultimately this led to the development of phase I trials of cisplatin in which impressive single-agent activity in patients with testicular cancer was seen. Unfortunately, in these trials, unique toxicity was also identified, including substantial emetogenic potential, nephrotoxicity, neurotoxicity, and ototoxicity.

In 1973, Dr Lawrence Einhorn and Dr John Donohue at Indiana University began the seminal study of cisplatin combination chemotherapy in patients with germ cell tumors. In the original study, patients were treated with cisplatin, vinblastine, and bleomycin (PVB) for 4 cycles of treatment followed by up to 2 years of maintenance therapy. The study also included the incorporation of aggressive postchemotherapy surgery for those who obtained only a partial remission with induction chemotherapy. The results of this Indiana University study changed forever the outlook and approach for patients with testicular cancer. Among the first 47 evaluable patients, 35 (74%) obtained a complete chemotherapy induced remission. An additional 12 patients (26%) obtained partial remission for an overall response rate of 100% and 5 of these 12 underwent aggressive postchemotherapy surgery to be rendered free of disease. Thirty-two patients (68%) obtained in durable remission. These overwhelming results were rapidly confirmed by other institutions in the United States and in Europe.

Subsequent to the initial demonstration of spectacular therapeutic effect of cisplatin combination chemotherapy, efforts turned to improving the tolerability of PVB, and such studies allowed the shaping of current modern therapy. Serial studies at Indiana University allowed two significant changes in the original PVB template. First, a randomized comparison of PVB using the original dose of vinblastine (0.4 mg/kg) to the same cisplatin and bleomycin dose but with a meaningful reduction of the vinblastine dose (0.3 mg/kg) demonstrated therapeutic equivalence and further confirmed the therapeutic punch of PVB in germ cell tumors. As expected, however, there were significant reductions in the myelotoxicity and neurotoxicity associated with the lower doses of vinblastine. Second, a randomized comparison of 4 courses of induction PVB (12 weeks) plus the original maintenance vinblastine schedule (up to 2 years) was compared with induction PVB alone (12 weeks). Again, the arms were therapeutically equivalent and induction therapy without maintenance became the standard approach (as it is today).

In the early 1980s, a new drug, etoposide, began to show activity in recurrent germ cell tumors. The toxicity profile of etoposide was primarily myelotoxicity and, thus, was able to be combined with cisplatin in full doses without overlapping toxicity. Approximately 25% of patients failing PVB responded to single-agent etoposide. Combinations of cisplatin and etoposide resulted in sustained remissions in some patients with recurrent germ cell tumors after PVB. The definitive contribution of etoposide in combination chemotherapy was demonstrated in the randomized comparison of 4 cycles of cisplatin and bleomycin with either vinblastine (PVB) or etoposide (BEP) as primary chemotherapy of disseminated germ cell tumors. The results reported by Dr Stephen Williams and colleagues from Indiana University and the Southeastern Cancer Study Group demonstrated the anticipated decrease in the neurotoxicity and musculoskeletal toxicity attributable to vinblastine. Moreover, PVB and BEP were at least therapeutically equivalent. In a randomized clinical trial in 261 men with disseminated germ cell tumors, etoposide was substituted for the vinblastine in this regimen to compare the efficacy and toxicity of the two treatments. Among 244 patients who could be evaluated for a response, 74% of those receiving PVB and 83% of those receiving BEP became disease-free with or without subsequent surgery but there was no difference in overall survival. Among those 157 patients classified as having high-volume disease, 61% became disease-free with PVB compared with 77% on BEP ( P <.05). Among the high–tumor volume patients, BEP was associated with higher overall survival ( P = .048). The regimens were similar in terms of myelosuppressive effects and pulmonary toxicity. The etoposide regimen, however, caused substantially fewer paresthesias ( P = .02), abdominal cramps ( P = .0008), and myalgias ( P = .00002).

As a sidelight to these therapeutic advances, careful clinical observations over this time period of rapid discovery of chemotherapy regimens for disseminated germ cell tumors defined the next era of clinical investigations wherein the effective standard regimens were tested and refined. The two categories of observations were development of prognostic indices to define pretreatment expectations of cure with cisplatin-based therapy and careful aggregation of short-term and long-term side-effect profiles to provide a comprehensive impact statement of curative chemotherapy.

Era of development (1985–2005)

Initial chemotherapy investigations led careful observers to a basic differentiation of patients with disseminated germ cell tumors. Even in early studies, it seemed as if there were a group of patients who had a 90% or greater chance of obtaining disease-free status and remaining in that status and a group for whom disease control was considerably less predictable with a 30% to 50% chance of failing to obtain disease-free status. Several institutions developed parochial prognostic systems that defined a good/minimal risk group, a poor/advanced risk group, and sometimes an intermediate-risk or moderate-risk group. This risk discrimination was usually made using readily available clinical parameters, such as anatomic extent of disease, site of primary, seminoma versus nonseminoma, and, in some systems, the degree of elevation of serum markers, such as α-fetoprotein (AFP), human chorionic gronadotropin (hCG), and lactate dehydrogenase (LDH). As can be imagined, these locally developed prognostic systems each had advantages and disadvantages and were variably reliable in predicting outcomes of systemic treatment. Nonetheless, the approximately ability to discriminate a good-risk group from a destined to do less well with standard cisplatin-based chemotherapy sent the field of clinical investigations for disseminated germ cell tumors down the path of testing toxicity reduction strategies in good-risk patients and examining approaches with enhanced therapeutic potential in those patients with predicted poor outcomes.

In good-risk germ cell tumors, the initial questions were aimed at duration of therapy of what was by then standard treatment of disseminated disease—BEP × 4. Investigators at Indiana University and the Southeastern Cancer Study Group performed a randomized clinical trial comparing the standard of 4 cycles of BEP to an experimental arm with the same doses and schedule of chemotherapy but given for 3 rather than 4 cycles. This trial included 184 patients with disseminated germ cell tumors in the Indiana University minimal-risk and moderate-risk class. In the minimal extent category, 106 of 107 patients became disease-free and overall 92% of patients receiving 3 cycles and 92% of patients receiving 4 cycles remained continuously disease-free. Among this group of 184 patients, there was only 1 therapy-related death: a single death related to neutropenic sepsis. The investigators concluded that 3 cycles of BEP was the preferred approach to patients with good-risk disease and the standard of care for good-risk disease changed to this 9-week regimen. These results were confirmed in long-term follow-up of this study. At 10 years, there were no differences in the 118 Indiana University patients receiving either 3 or 4 cycles of BEP. Incorporating serum markers did suggest, however, that patients with hCG greater than 1000 mIU/mL were at higher risk of failure (5/14 deaths) compared with those with hCG less than 1000 mIU/mL (2/104).

In a final attempt to minimize therapy in good-risk patients, investigators at Indiana University and the Eastern Cooperative Oncology Group conducted a randomized trial of 3 cycles of BEP (the new standard for favorable prognosis patients) to 3 cycles of cisplatin and etoposide (EP) in the same dose and schedule as BEP but with the deletion of what was felt to be potentially the most vexing component of the regimen: bleomycin. This trial has defined the floor of minimizing therapy for good-risk disease. Including both Indiana University minimal and moderate disease categories, 178 patients were randomized. There were no clinically significant pulmonary toxicity on either arm and the single therapy-related death was related to sepsis while on BEP. There was a 14% incidence of grade IV neutropenia on the BEP versus 7% on EP. Overall, 94% of patients who received BEP and 88% who received EP achieved a disease-free status with chemotherapy and/or surgery. There was significant difference in treatment failures, however, including persistent carcinoma in postchemotherapy resections of residual disease and relapses from complete remission, which occurred on the arm without bleomycin. Also, a significant negative impact on failure-free (86% vs 69%; P = .01) and overall survival (95% vs 86%; P = .01) was demonstrated for those receiving EP without bleomycin. The investigators concluded that bleomycin is an essential component of cisplatin combination chemotherapy in patients who receive 3 cycles of treatment for minimal-stage or moderate-stage disseminated germ cell tumors.

Other investigators took a different path trying to refine therapy for good-risk disease based on attempts to reduce the chemotherapy toxicity profile related to cisplatin by substituting carboplatin and reducing the schedule and cumulative dose of bleomycin as well as using attempts to shorten the duration of the traditional 5-day treatments to 3 days. In sum, none of these well-conceived efforts resulted in superior therapeutic profiles or clinically significant improvement in the acute toxicity profile while retaining therapeutic equivalence to BEP × 3.

There have been investigations comparing BEP × 4 with EP × 4 and BEP × 3 with EP × 4 in favorable prognosis patients to test the relative therapeutic equivalence and comparative toxicities. deWit and colleagues with the European Organisation for Research and Treatment of Cancer (EORTC) conducted a randomized trial of 4 cycles of BEP compared with 4 cycles of EP in favorable prognosis patients. Patients were categorized by the local prognostic system at the time, not the International Germ Cell Consensus Classification (IGCCC), but most patients fell into the Indiana University minimal disease classification; 419 patients were randomized to receive either BEP × 4 or EP × 4. Patients received a nonstandard dose of etoposide (120 mg/m ² , days 1, 3, and 5 of the regimen) compared with American BEP (100 mg/m ² daily × 5). With treatment, 87% of patients allocated to EP × 4 achieved complete response compared with 95% of those allocated to BEP × 4 ( P = .0075). At 7.3 years’ median follow-up, 8 patients in each arm (4%) relapsed. Given the low incidence of unfavorable outcomes, no significant differences could be detected in progression-free or overall survival. In this schedule of 4 cycles of therapy, there were significant differences in neurotoxicity and acute pulmonary toxicity. Two patients on the BEP × 4 arm died with death attributable to bleomycin. Overall, in this large randomized trial in favorable patients with germ cell tumor, it was concluded that bleomycin was associated with additional toxicity but was an essential component that could not be deleted without sacrificing favorable therapeutic outcomes.

Finally, Culine and colleagues in the French Federation of Cancer Centers Genito-Urinary Group conducted the logical concluding trial in this sequence, which compared the standard therapy for good-risk germ cell tumors, BEP × 3, with the 4-cycle regimen popularized at Memorial Sloan-Kettering, EP × 4. With the exception of deletion of bleomycin in the EP × 4, dose and schedule were American BEP, with bleomycin (30 units, days, 1, 8, and 15) with cisplatin (20 mg/m ² daily, days 1–5) and etoposide (100 mg/m ² daily, days 1–5). This trial was powered for equivalence and again the initial prognostic system used was the prevailing regional system and not the IGCCC; 257 eligible patients were randomized and analyzed. There was no statistical difference in obtaining complete remission by arm (BEP ×3:94% vs EP ×4:96%). Twenty patients relapsed (6 on BEP × 3 vs 14 on EP × 4) and 17 died (5 BEP × 3 vs 12 EP × 4). There were no treatment-related deaths. There was no statistical difference in pulmonary toxicity. The trial was not powered to examine event-free survival or overall survival. When these parameters were analyzed by either the prevailing prognostic system at the time of the study or after re-analysis using standardized the IGCCCC, no significant trends favoring BEP × 3 were shown. The authors and most world experts have interpreted this trial as definitive, and the risk of jeopardizing a large group of good-risk patients to what may well be substandard, equitoxic therapy to answer this question statistically was not felt to be merited. The primacy of BEP × 3 as standard therapy for IGCCC good-risk patients was upheld and has now been endorsed by emerging guidelines worldwide.

Era of development (1985–2005)

Initial chemotherapy investigations led careful observers to a basic differentiation of patients with disseminated germ cell tumors. Even in early studies, it seemed as if there were a group of patients who had a 90% or greater chance of obtaining disease-free status and remaining in that status and a group for whom disease control was considerably less predictable with a 30% to 50% chance of failing to obtain disease-free status. Several institutions developed parochial prognostic systems that defined a good/minimal risk group, a poor/advanced risk group, and sometimes an intermediate-risk or moderate-risk group. This risk discrimination was usually made using readily available clinical parameters, such as anatomic extent of disease, site of primary, seminoma versus nonseminoma, and, in some systems, the degree of elevation of serum markers, such as α-fetoprotein (AFP), human chorionic gronadotropin (hCG), and lactate dehydrogenase (LDH). As can be imagined, these locally developed prognostic systems each had advantages and disadvantages and were variably reliable in predicting outcomes of systemic treatment. Nonetheless, the approximately ability to discriminate a good-risk group from a destined to do less well with standard cisplatin-based chemotherapy sent the field of clinical investigations for disseminated germ cell tumors down the path of testing toxicity reduction strategies in good-risk patients and examining approaches with enhanced therapeutic potential in those patients with predicted poor outcomes.

In good-risk germ cell tumors, the initial questions were aimed at duration of therapy of what was by then standard treatment of disseminated disease—BEP × 4. Investigators at Indiana University and the Southeastern Cancer Study Group performed a randomized clinical trial comparing the standard of 4 cycles of BEP to an experimental arm with the same doses and schedule of chemotherapy but given for 3 rather than 4 cycles. This trial included 184 patients with disseminated germ cell tumors in the Indiana University minimal-risk and moderate-risk class. In the minimal extent category, 106 of 107 patients became disease-free and overall 92% of patients receiving 3 cycles and 92% of patients receiving 4 cycles remained continuously disease-free. Among this group of 184 patients, there was only 1 therapy-related death: a single death related to neutropenic sepsis. The investigators concluded that 3 cycles of BEP was the preferred approach to patients with good-risk disease and the standard of care for good-risk disease changed to this 9-week regimen. These results were confirmed in long-term follow-up of this study. At 10 years, there were no differences in the 118 Indiana University patients receiving either 3 or 4 cycles of BEP. Incorporating serum markers did suggest, however, that patients with hCG greater than 1000 mIU/mL were at higher risk of failure (5/14 deaths) compared with those with hCG less than 1000 mIU/mL (2/104).

In a final attempt to minimize therapy in good-risk patients, investigators at Indiana University and the Eastern Cooperative Oncology Group conducted a randomized trial of 3 cycles of BEP (the new standard for favorable prognosis patients) to 3 cycles of cisplatin and etoposide (EP) in the same dose and schedule as BEP but with the deletion of what was felt to be potentially the most vexing component of the regimen: bleomycin. This trial has defined the floor of minimizing therapy for good-risk disease. Including both Indiana University minimal and moderate disease categories, 178 patients were randomized. There were no clinically significant pulmonary toxicity on either arm and the single therapy-related death was related to sepsis while on BEP. There was a 14% incidence of grade IV neutropenia on the BEP versus 7% on EP. Overall, 94% of patients who received BEP and 88% who received EP achieved a disease-free status with chemotherapy and/or surgery. There was significant difference in treatment failures, however, including persistent carcinoma in postchemotherapy resections of residual disease and relapses from complete remission, which occurred on the arm without bleomycin. Also, a significant negative impact on failure-free (86% vs 69%; P = .01) and overall survival (95% vs 86%; P = .01) was demonstrated for those receiving EP without bleomycin. The investigators concluded that bleomycin is an essential component of cisplatin combination chemotherapy in patients who receive 3 cycles of treatment for minimal-stage or moderate-stage disseminated germ cell tumors.

Other investigators took a different path trying to refine therapy for good-risk disease based on attempts to reduce the chemotherapy toxicity profile related to cisplatin by substituting carboplatin and reducing the schedule and cumulative dose of bleomycin as well as using attempts to shorten the duration of the traditional 5-day treatments to 3 days. In sum, none of these well-conceived efforts resulted in superior therapeutic profiles or clinically significant improvement in the acute toxicity profile while retaining therapeutic equivalence to BEP × 3.

There have been investigations comparing BEP × 4 with EP × 4 and BEP × 3 with EP × 4 in favorable prognosis patients to test the relative therapeutic equivalence and comparative toxicities. deWit and colleagues with the European Organisation for Research and Treatment of Cancer (EORTC) conducted a randomized trial of 4 cycles of BEP compared with 4 cycles of EP in favorable prognosis patients. Patients were categorized by the local prognostic system at the time, not the International Germ Cell Consensus Classification (IGCCC), but most patients fell into the Indiana University minimal disease classification; 419 patients were randomized to receive either BEP × 4 or EP × 4. Patients received a nonstandard dose of etoposide (120 mg/m ² , days 1, 3, and 5 of the regimen) compared with American BEP (100 mg/m ² daily × 5). With treatment, 87% of patients allocated to EP × 4 achieved complete response compared with 95% of those allocated to BEP × 4 ( P = .0075). At 7.3 years’ median follow-up, 8 patients in each arm (4%) relapsed. Given the low incidence of unfavorable outcomes, no significant differences could be detected in progression-free or overall survival. In this schedule of 4 cycles of therapy, there were significant differences in neurotoxicity and acute pulmonary toxicity. Two patients on the BEP × 4 arm died with death attributable to bleomycin. Overall, in this large randomized trial in favorable patients with germ cell tumor, it was concluded that bleomycin was associated with additional toxicity but was an essential component that could not be deleted without sacrificing favorable therapeutic outcomes.

Finally, Culine and colleagues in the French Federation of Cancer Centers Genito-Urinary Group conducted the logical concluding trial in this sequence, which compared the standard therapy for good-risk germ cell tumors, BEP × 3, with the 4-cycle regimen popularized at Memorial Sloan-Kettering, EP × 4. With the exception of deletion of bleomycin in the EP × 4, dose and schedule were American BEP, with bleomycin (30 units, days, 1, 8, and 15) with cisplatin (20 mg/m ² daily, days 1–5) and etoposide (100 mg/m ² daily, days 1–5). This trial was powered for equivalence and again the initial prognostic system used was the prevailing regional system and not the IGCCC; 257 eligible patients were randomized and analyzed. There was no statistical difference in obtaining complete remission by arm (BEP ×3:94% vs EP ×4:96%). Twenty patients relapsed (6 on BEP × 3 vs 14 on EP × 4) and 17 died (5 BEP × 3 vs 12 EP × 4). There were no treatment-related deaths. There was no statistical difference in pulmonary toxicity. The trial was not powered to examine event-free survival or overall survival. When these parameters were analyzed by either the prevailing prognostic system at the time of the study or after re-analysis using standardized the IGCCCC, no significant trends favoring BEP × 3 were shown. The authors and most world experts have interpreted this trial as definitive, and the risk of jeopardizing a large group of good-risk patients to what may well be substandard, equitoxic therapy to answer this question statistically was not felt to be merited. The primacy of BEP × 3 as standard therapy for IGCCC good-risk patients was upheld and has now been endorsed by emerging guidelines worldwide.