Hepatoblastoma, like several other solid tumors in children, is a good-news story: good news for the spectacular improvement in survival for hepatoblastoma achieved over the past 30 to 40 years and also good news for the manner in which this was achieved—the remarkable level of collaboration by several international cooperative clinical trials groups. Hepatoblastoma is a rare malignant embryonal tumor of the liver that occurs in young infants with a median age at diagnosis of 16 months.1 With an age-adjusted frequency of 1.2 per million children aged <15 years, hepatoblastoma accounts for <1% of new cancer diagnoses in childhood.2 Ironically, it is the very rarity of hepatoblastoma that may have contributed to its improved clinical outcome. Most pediatric oncology units, which treat between 50 and 150 new cases of childhood cancer per year, will only diagnose 1 new case of hepatoblastoma per year or even less. However, the treatment of a malignant liver tumor in a young infant requires a special level of interdisciplinary collaboration between oncologists, diagnostic radiology and pathology, and surgeons to ensure the best possible outcome. Seeking an improved outcome fostered the development of the first collaborative clinical trials for infants with hepatoblastoma and, in doing so, has provided healthcare professionals with a clear roadmap for the care of a rare malignant tumor based on the best available evidence.
In 1982, Evans et al reported what to our knowledge was the first collaborative trials for childhood liver tumors. After the first chemotherapy trial, in which a 17% survival rate was achieved, a second study of 62 patients (Children's Cancer Group [CCG]/Pediatric Oncology Group [POG], 1976-1978) who received postoperative vincristine, doxorubicin, cyclophosphamide, and 5-flurouracil achieved a 30% survival rate and demonstrated for the first time that hepatoblastoma is a chemosensitive tumor.3 A later nonrandomized CCG trial (CCG 823. 1986-1989) used cisplatin and doxorubicin (Adriamycin) for unresectable and metastatic hepatoblastoma and achieved 3-year event-free survival (EFS) rates of 55% and 30%, respectively.4 A similar study from the POG that used cisplatin, 5-fluorouracil, and vincristine (C5V) achieved a similar 3-year EFS rate.5
Although later treatment approaches varied, subsequent cooperative clinical trials from the Intergroup Hepatoma Study, the German Society of Pediatric Oncology and Hematology (GPOH), and the International Society of Pediatric Oncology (SIOP) Liver Tumor Study Group (SIOPEL) all produced dramatic improvements in overall survival for hepatoblastoma; and central to that success was the use chemotherapy regimens that incorporated cisplatin. The Intergroup Study INT-0098 (1989-1992) of postoperative C5V or combined cisplatin and doxorubicin produced a 3-year overall survival rate of 71%; and a similar rate was produced in the German GPOH HB99 study, which used postoperative chemotherapy with combined ifosfamide, cisplatin, and doxorubicin.6, 7
The first SIOP Liver Tumor trial, SIOPEL 1 (1990-1994), enrolled 155 patients to receive preoperative treatment with cisplatin and doxorubicin and achieved a 5-year overall survival rate of 75%.1 In SIOPEL 2, 1 arm of a risk-adapted strategy evaluated single-agent cisplatin given preoperatively for standard-risk tumors and achieved an impressive 3-year overall survival rate of 91%.8 The change from a largely fatal malignant tumor to a mostly curable tumor using single-agent cisplatin and expert surgery is a remarkable outcome.
However, cure comes at a cost. Cisplatin, which is used widely in the treatment of childhood solid tumors (notably, neuroblastoma, germ cell tumors, osteosarcoma, medulloblastoma, and hepatoblastoma), causes measurable hearing loss in 25% to 90% of children.9
Risk factors for ototoxicity include preexisting hearing loss, young age, cumulative cisplatin dose (>300 mg/m2), dose schedule, coexisting renal dysfunction, and prior cranial radiotherapy, especially when the cochlea is within the radiation field.9, 10 Hearing loss is particularly marked in young children, especially those aged <5 years, who have a relative risk of significant hearing loss that is 20-fold greater than that for an individual aged 20 years who receives the same cisplatin dose.10 Moreover, for very young, preverbal infants, the impact of measurable hearing loss maybe greatly underestimated, because cisplatin has its greatest impact in the high and very high frequency ranges of hearing: the same frequency range that very young children depend on as they acquire language, develop social skills, and enter structured learning environments.9
However, not all children of the same age who receive the same cumulative cisplatin dose experience hearing loss, suggesting that genetic factors are involved. Indeed, the risk of cisplatin hearing impairment is associated with specific genotypes of glutathione S-transferase.11
Other platins (notably, carboplatin) are less ototoxic than cisplatin but maybe less effective; therefore, despite the high risk of ototoxicity, cisplatin remains the agent of choice for hepatoblastoma and many other childhood solid tumors for the foreseeable future. Seeking to prevent a lifetime of treatment-related deafness, several recent trials have evaluated new agents aimed at reducing ototoxicity.
In this issue, Katzenstein et al report the hearing outcomes of a randomized trial of amifostine for reducing platin-associated toxicity, most notably cisplatin-related ototoxicity. In their Intergroup Hepatoblastoma study (P9645),12 120 patients were randomized to treatment, depending on disease stage, to either 4 cycles of C5V or 6 cycles of cisplatin and carboplatin (cumulative cisplatin dose of 400-600 mg/m2), with or without amifostine. Available data for 82 randomized patients were evaluated for hearing loss using modified Brock criteria.
Amifostine is an aminothiol prodrug that is dephosphorylated rapidly to an active thiol. In normal tissues, it acts as a reactive oxygen scavenger and also covalently binds cisplatin.13, 14 Originally developed as a radiation protectant, encouraging data from animal studies led to several clinical trials in adults and children to assess its effectiveness in preventing cisplatin ototoxicity and nephrotoxicity.14
Although several early adult studies demonstrated efficacy against platin-related toxicity, in their current study, Katzenstein et al demonstrate that amifostine did not protect against any level of detectable hearing loss compared with controls. This result is in keeping with several smaller studies in children in which amifostine did not prevent cisplatin-related hearing loss in osteosarcoma, neuroblastoma, or germ cell tumors.15-17 However, in a recent study of 97 children with medulloblastoma, Fouladi et al demonstrated that amifostine significantly reduced severe hearing loss in patients who received craniospinal radiotherapy and dose-intense, cisplatin-based chemotherapy.18
However, several crucial differences are apparent between these 2 recent amifostine trials. First, the cumulative amifostine dosing and dose schedules differed: Katzenstein et al used a single dose of amifostine immediately before cisplatin in their hepatoblastoma study, whereas Fouladi et al gave an additional bolus dose of amifostine after 3 hours of the cisplatin infusion to their patients with medulloblastoma. Second, all patients with medulloblastoma received craniospinal radiotherapy, which itself is ototoxic and also is a known contributor to cisplatin-related hearing loss, whereas patients in the hepatoblastoma study (and in other solid tumor studies) received cisplatin-based chemotherapy alone. Third, the patients who had hepatoblastoma were significantly younger (median age, 1 year) compared with the patients who had medulloblastoma (median age, 8 years). and younger age carries a greater risk of more severe cisplatin-related hearing loss. Finally, the assessment of hearing and the classification of hearing loss were not comparable between the 2 studies.
Although amifostine may play a role in reducing ototoxicity in older children who are treated with cisplatin and craniospinal radiotherapy, it has no benefit in preventing cisplatin hearing loss in the very young, and those patients are at the greatest risk of hearing loss and in greatest need of preserving their hearing for their future development. However, several newer agents that currently are entering clinical trials may reduce cisplatin-related toxicity, and specifically ototoxicity.13, 14
Cisplatin ototoxicity results from the production of reactive oxygen species within the cochlea, overwhelming endogenous antioxidant mechanisms and causing irreversible free–radical-related apoptosis of cochlea outer hair cells, spiral ganglion cells, and the stria vascularis.14 In animal studies, cisplatin exposure to cochlea outer hair cell induces the expression of nicotinamide adenine dinucleotide phosphatase (NADPH) oxidase 3, which catalyzes superoxide synthesis and activates the intrinsic mitochondrial apoptosis cascade, resulting in caspase 3-induced and caspase 9-induced cell death. In the cochlea lateral wall and stria vascularis, cisplatin exposure leads to the activation of nuclear factor κB and the formation of nitric oxide (NO) by inducible NO synthase (iNOS) with the induction of caspase 3-related apoptosis. While in the spiral ganglion, cisplatin induces high mobility group 1 protein expression, which also leads to iNOS-related apoptosis.13, 14, 19
In principle, preventing cisplatin-related hearing loss requires agents that are capable of selectively protecting cochlea cells from apoptotic cell death while not preventing tumor-related apoptosis. For practical reasons, local chemoprotectant therapy administered directly into the inner ear is unlikely to be used in very young children.
However, several systemic agents have been identified recently as candidates for preventing cisplatin-related ototoxicity. The most promising is sodium thiosulfate (STS), which, similar to amifostine, is a free–radical-scavenging thiol agent.19 In vitro, STS acts in several ways: it directly inactivates cisplatin by the covalent binding of cisplatin to its thiol moiety to form an inactive complex, it scavenges cisplatin-related reactive oxygen species, and it may concentrate in the perilymph or endolymph, further inactivating cisplatin in the inner ear.14, 19, 20
Recently, SIOPEL (including the German GPOH, the Japanese Study Group for Pediatric Liver Tumors, and several US centers) opened the first large randomized trial of STS (SIOPEL 6) to reduce the burden of ototoxicity in children with hepatoblastoma. That phase 3 trial will enroll 250 children with standard-risk hepatoblastoma for preoperative treatment with single-agent cisplatin or cisplatin and STS, and the STS will be given as a delayed infusion 6 hours after cisplatin infusion to rescue cisplatin ototoxicity without compromising antitumor activity. A standardized approach to the evaluation of ototoxicity has been developed for the study. Secondary objectives of SIOPEL 6 will evaluate nephrotoxicity and investigate genetic loci associated with cisplatin ototoxicity. In a parallel study, the Children's Oncology Group recently activated a randomized phase 3 clinical trial (ACCL0431) to evaluate the efficacy of STS for preventing cisplatin-related hearing loss in newly diagnosed children with hepatoblastoma, germ cell tumors, medulloblastoma, and osteosarcoma.21
However, even the most thoroughly conducted and controlled collaborative studies are only as good as the data available to report. In the study reported here by Katzenstein et al, 120 patients with hepatoblastoma made up the evaluation cohort, but >30% of those patients (n = 38) had to be excluded from the analysis because of incomplete audiologic or clinical data. There is a need for future ototoxicity trials to use common criteria for the assessment and reporting of hearing loss in young children, as Katzenstein et al observe.
This represents a significant challenge for the design and comparison of future international collaborative studies in which toxicity reduction is the measurable endpoint. There is a message in this. In the past, when the goal was to assess treatment response and improve survival, data quality has rarely been an issue. However, the goalposts now have shifted to enhancing long-term survival, and with this comes the need to ensure that the study design and measurement of long-term toxicity, such as hearing loss, become equally valued objectives.