The prognosis of children who are affected by hepatoblastoma (HB) that presents with lung metastases has always been considered very poor. In light of the overall improvement in the survival of HB patients since the introduction of cisplatin (CDDP) in the therapeutic armament of this tumor, the question has been raised whether patients with metastatic HB also would benefit from this drug. The purpose of the current study was to address this issue by analyzing the treatment outcome of those patients presenting with metastases who entered into the first HB study on childhood liver tumors conducted by the International Society of Paediatric Oncology (SIOPEL 1).
SIOPEL 1 was a prospective, international, multicentric, single-arm study based on preoperative chemotherapy that was open to patient registration from January 1990 to February 1994. After undergoing a biopsy, patients received four courses of CDDP (80 mg/m2 in a 24-hour, continuous infusion) on Day 1 followed by doxorubicin (60 mg/m2 in a 48-hour, continuous infusion) on Days 2 and 3 (PLADO). Surgery was performed after four courses of PLADO and was followed by two more courses. Untreated children age < 16 years with biopsy-proven HB were eligible for the study. Metastatic spread was assessed by chest X-ray and, where available, lung computed tomography scan.
Thirty-one of 154 children that entered into the trial presented with metastases. Eight children presently are alive with no evidence of disease (NED) after being treated with protocol therapy only (median follow-up, 60 months); nine children are alive with NED after having failed PLADO and having been rescued with alternative therapies (median follow-up, 80 months). The 5-year overall and event free survival rates for these children were 57% (95% confidence interval, 39–75%) and 28% (95% confidence interval, 12–44%), respectively. Persistent lung disease was the main reason for PLADO failure (17 of 23 patients; 74%).
The introduction of cisplatin (CDDP) into the therapeutic armamentarium of childhood hepatoblastoma (HB) has significantly improved the prognosis of children affected by this tumor. The 3-year overall survival (OS) rate of these patients has increased from 30% in the late 1970s1 to 60–70% reported by the main, modern, CDDP-based therapeutic trials.2–5 However, it remains to be determined which patients benefited most from the new CDDP-containing regimens. In the first cooperative trial on childhood HB run by the International Society of Paediatric Oncology (SIOP) in the early 1990s—SIOPEL 1—the presence of lung metastases at the time of diagnosis negatively influenced patient outcome.6 For almost all malignant tumor types, the presence of metastases implies a poor prognosis or, at the very least, a prognosis that is worse compared with patients who present with localized disease. Particularly for HB patients, before the CDDP era, only anecdotal cases of long term survivors among children with metastases were reported.1, 7 Thus, it becomes important to investigate whether, despite what is known already about the dismal prognosis of these children, the modern therapeutic strategies improve these historic data.
The objective of this report was to analyze in detail the results obtained with the therapeutic strategy, adopted for the SIOPEL 1 study, in the cohort of HB patients who presented with (lung) metastases who entered the trial. The natural history of this disease under the influence of the modern CDDP-based therapeutic standards also is analyzed in an effort to identify possible ways of improving the present results.
MATERIALS AND METHODS
The SIOPEL 1 trial was an international, prospective, cooperative, single-arm study that was open for patient registration between January 1990 and February 1994 and involved 91 centers from 30 countries. Children age < 16 years with a previously untreated, histologically proven HB were eligible for the study. Central review of the pathology slides was required for each patient entering the study.
Required Patient Information at Diagnosis and Staging Procedures
An accurate physical examination, including anthropometric and nutritional status evaluation, an assessment of hepatic and renal function, determination of the serum α-fetoprotein (α-FP) concentration value, and an echocardiogram measurement of the left ventricular ejection fraction and shortening fraction were requested prior to starting treatment. Peripheral blood was analyzed for a full blood count and electrolyte levels, including magnesium and creatinine. Open or closed surgical biopsy was required for diagnosis. All material obtained at biopsy and resection was reviewed centrally by one pathologist (J.W.K.). Informed consent was obtained from patients and/or guardians before instituting therapy and always included Research Ethics Committee approval.
Staging Procedures and the Pretreatment Extent of Disease System
The pretreatment assessment of the extent of the primary tumor was by abdominal ultrasound, and/or computed tomography (CT) scan, and/or magnetic resonance imaging. Metastatic spread was assessed by chest X ray (posteroanterior and lateral) and, where available, lung CT scan. Hepatic angiography was optional. Central review of imaging was not performed. Based on the radiologic findings, a pretreatment extent of disease (PRETEXT) system was developed to describe tumor extent at the time of diagnosis (before any therapeutic act) and during therapy (Fig. 1). In the PRETEXT system, the liver is divided into four sections: anterior and posterior sections on the right and medial and lateral sections on the left. In this way, four PRETEXT categories are identified: PRETEXT I, in which three adjoining sections are free and one section is involved by the tumor; PRETEXT II, in which two adjoining sections are free and two sections are involved; PRETEXT III, in which just one section is free and three sections are involved (or two nonadjoining sections are tumor free); and PRETEXT IV, in which there are no tumor free sections.
Extension of the tumor beyond the liver was indicated with “V” when there was extension into the vena cava and/or all three hepatic veins, with “P” when there was extension into the main and/or both left and right branches of the portal vein, with “E” when there was extrahepatic extension (except for P and V), and with “M” when there was the presence of distant metastases. Thus, the definitive tumor extension is expressed in terms of PRETEXT category (I–IV) and presence or absence of extrahepatic disease (V, P, E, or M).
Treatment Plan and Chemotherapy Details
All patients (regardless of any clinical or tumor characteristics) were to receive the same treatment, which consisted of a preoperative combination of CDDP and doxorubicin (DOXO) chemotherapy (PLADO; see below) every 3 weeks and, where feasible, resection of the primary tumor after four courses of PLADO. Each course of PLADO chemotherapy was comprised of CDDP on Day 1 at a dose of 80 mg/m2 in a continuous, 24-hour infusion and DOXO at doses of 30 mg/m2 per day given as a continuous, 24-hour, intravenous infusion on Days 2 and 3 (total dose per course, 60 mg/m2). After surgery, two more courses of PLADO were administered (starting within the first 2 weeks after the operation). No more than six courses of PLADO were to be given to any patient. Thus, the total planned doses were 480 mg/m2 of CDDP and 360 mg/m2 of DOXO. The objective of limiting the total dose of both drugs was to reduce the incidence of cardiotoxicity from DOXO and nephropathy and ototoxicity from CDDP. The doses recommended for children who weighed < 10 kg were calculated per kilogram. In the case of hepatic dysfunction and/or serum bilirubin concentration > 3 mg/100 mL and liver enzymes (aspartate aminotransferase and alanine aminotransferase) > 5 times the normal value, the DOXO dose was to be reduced by 50%. Courses of PLADO were repeated every 21 days if the absolute neutrophil count was > 1.0 × 109/L and the platelet count was > 100 × 109/L. Older children received pre-CDDP hydration at a rate of 125 mL/kg per hour. The pre-CDDP infusion contained 10 mmol of potassium chloride, 2 mmol of magnesium sulphate, and 1.5 mmol of calcium as calcium chloride per 500 mL of glucose 2.5%, with sodium chloride 0.45% to be infused at 125 mL/m2 per hour for 12 consecutive hours. During and after CDDP administration, intravenous fluids were given at the same rate but with an additional 30 mL of 20% mannitol solution added to every 500 mL of the glucose-electrolyte solution. DOXO was administered as a continuous infusion through a central venous catheter. If the urine output fell below 400 ml/m2 in 6 hours, then an intravenous dose of furosemide was advised. The use of nephrotoxic antibiotics was discouraged during and immediately after the CCDP infusion. Plasma electrolytes were to be monitored carefully throughout chemotherapy, particularly in infants.
If the primary tumor could not be resected by partial hepatectomy, then liver transplantation could be considered, provided there had been a response to PLADO chemotherapy and there was no evidence of extrahepatic disease. No specific guidelines were provided for patients with lung metastases, except for recommending lung surgery, if feasible, in patients with persisting pulmonary lesion(s) after PLADO therapy; however, whole lung irradiation was not recommended. Below, the entire SIOPEL 1 treatment strategy is referred to as the protocol therapy.
Timing of Tumor Response Evaluation and Surgery
Tumor response was evaluated after the second and fourth courses of PLADO (by repeating the positive instrumental examinations of diagnosis and monitoring the serum α-FP level). If there was evidence of progressive disease at either time point, as defined below, then the patient was switched to alternative chemotherapy. After four courses of PLADO, tumor resectability was assessed. If feasible, partial hepatectomy was then performed, and two more courses of PLADO were given postoperatively. If the tumor was responding to chemotherapy but was still considered unresectable, then two more courses of PLADO were given, and tumor resectability was assessed again. Thus, as outlined above, no more than six courses of PLADO were to be administered to any patient.
Tumor response was defined as a complete response if there was no evidence of disease (NED) and the serum α-FP level was normal, a partial response if there was any tumor volume shrinkage associated with a decreasing α-FP level, stable disease if there was no change in tumor volume and in the serum α-FP level, and progressive disease if there was an unequivocal increase in one or more dimensions and/or any unequivocal increase of the serum α-FP level (three successive 1–2 weekly determinations) even without clinical evidence of tumor regrowth. The protocol required separate evaluation of response for the primary tumor and each metastasis. The overall response was derived from the “worst-responding” lesion.
Monitoring of Toxicity
Prior to each course of chemotherapy, physical examination, full blood count, urea and electrolytes, liver function tests, and serum α-FP level were required. Renal, cardiac, and auditory functions were assessed after alternate courses of PLADO. Renal function was monitored by measurement of the glomerular filtration rate.
The results of this trial are expressed in terms of 1) response to PLADO, 2) OS, and 3) event free survival (EFS). All calculations are based on “intention to treat” and include data on patients who died before any treatment could be given or during the first weeks of chemotherapy. To be included in the response rate calculation, patients must have received at least two courses of chemotherapy. OS was defined as the time interval between the date of diagnosis and the date of death (from any cause) or the date of last follow-up. EFS was defined as the time interval from the date of diagnosis to 1) the date of alternative treatment because of failure of the PLADO regimen, 2) the date of disease recurrence, 3) the date of death, or 4) the date of last follow-up, whichever occurred first. The date of alternative treatment was used as a surrogate for the date of disease progression when the precise date of progression was not recorded. A disease recurrence was identified from unequivocal imaging and serial elevated serum α-FP levels or by biopsy confirmation. Patients who were lost to follow-up were censored in these analyses. The Kaplan–Meier method was used to estimate survival curves.8 Ninety-five percent confidence intervals (95% CIs) were calculated for the survival estimates using the Greenwood method.9 All statistical procedures were carried out using the SAS statistical package (SAS, Inc., Cary, NC).
Thirty-one of the 154 children (20%) who entered the trial during the study period presented with lung metastases; in 1 child, at the time of the initial minilaparotomy obtaining the histologic diagnosis, multiple peritoneal nodules also were visible (not otherwise seen in the abdominal CT scan). The main clinical characteristics of these patients are reported in Table 1. The distribution of patients in this cohort by age, gender, α-FP level at the time of diagnosis and PRETEXT category were similar to the distribution observed in children with localized HB (data not shown). All children underwent a diagnostic biopsy, and the diagnosis was confirmed centrally by one of the authors (J.W.K.).
Table 1. Main Clinical Characteristics of the 31 Patients who Presented with Lung Metastases and Entered Into the First Study of the International Society of Paediatric Oncology on Childhood Liver Tumors
Tweve children (39%) who completed therapy according to the protocol guidelines were considered disease free (i.e., there was no evidence of measurable disease and the patient had a negative serum α-FP level) at the time that they stopped therapy and, thus, went off therapy (Fig. 2). Eight of these 12 patients (26% of the study population) are presently alive with NED at a median follow-up from the time of diagnosis of 60 months. None of them had thoracotomies to remove residual lung disease. One child with PRETEXT IV disease exhibited lungs free of metastases after four courses of PLADO; however, due to an unresectable primary tumor, after two more courses of PLADO, the patient underwent an orthotopic liver transplantation (OLT). The child is alive more than 41 months from diagnosis and more than 34 months from OLT.
Four of 12 children who went off therapy eventually suffered a tumor recurrence: 3 patients in the lungs at 8 months, 15 months, and 51 months from the time of diagnosis and 1 patient locally at 34 months. At their last follow-up, three of these four children are alive NED at more than 83 months, 65 months, and 60 months from the time of diagnosis and at more than 66 months, 40 months, and 31 months from the time of disease recurrence, respectively, after having been saved with further chemotherapy and thoracotomies. The remaining patient (who had a lung recurrence) is alive with evidence of disease at more than 51 months from the time of diagnosis.
Children who were never considered disease free
Ninteen children (61%) were never considered disease free either because of evidence of measurable disease (18 children) or a persisting elevated α-FP level (1 patient) at the end of protocol therapy. One child died of disease after one course of PLADO from internal bleeding nine days after starting therapy. Among the remaining 18 children, 14 had a response to PLADO, with complete clearing of the lung deposits in 5 of 14 patients and 4 suffered of progressive disease.
Of the 5 patients with lungs that were cleared by the protocol therapy, 2 died of disease at 15 months and 22 months from the time of diagnosis, respectively, and 3 patients were censored alive with NED at more than 90 months for two of the 3 patients, and more than 80 months for 1 patient from the time of diagnosis at their last follow-up. Two of these latter 3 patients, who received 2 months and 3 months, respectively, of further chemotherapy after protocol therapy, also had an OLT (present follow-up from OLT: 75+ months and 70+ months, respectively).
Among the remaining 9 patients with tumors that responded to PLADO but with lungs that never cleared, 2 children are alive with NED at more than 91 months and more than 41 months from the time of diagnosis after further chemotherapy and surgery, 2 children died as a result of surgical complications, and 5 children died of disease at 16 months, 20 months, 21 months, 49 months, and 54 months from the time of diagnosis.
In four patients, the tumor progressed during PLADO chemotherapy. Three of these patients died of disease at 2 months, 2 months, and 9 months from the time of diagnosis. It is noteworthy that one child was alive with NED at 54 months before having been lost to follow-up. After PLADO failure, this child was treated with carboplatin and etoposide. The tumor responded, the lungs cleared, and the primary tumor became resectable.
Overview of the Treatment Results
Twenty-six children had a partial response to PLADO (84%; 95% CI, 69–92%) with complete clearing of lung metastases in 17 patients (55%; 95% CI, 52–68%). However, only 8 of these 17 patients were maintained in complete remission of the pulmonary disease with PLADO alone; they represent 26% of the children (8 of 31 patients) with metastatic HB who entered the trial and 47% of the children (8 of 17 patients) who had a complete remission of the lung deposits on PLADO therapy.
Twenty-three children failed the protocol therapy (74%), and 17 children (74%) did so because of the lung disease. Thirteen patients died (42%): 1 from tumor bleeding soon after the first course of chemotherapy, 2 from surgical complications, and 10 from disease at a median time interval from diagnosis to death of 18 months (range, 2–54 months). In 9 of these 13 children who died, the lungs never cleared (9 of 13 patients; 69%).
In total, 17 children (55%; 17 of 31 patients) are alive with NED with a median follow-up from the time of diagnosis of 58.5 months (range, 36–83 months). This means that other nonprotocol therapies were capable of saving 9 children for whom PLADO was not curative (median follow-up, 80 months). Four of these nine long term survivors underwent a single thoracotomy (two patients) or multiple thoracotamies (two surgeries in one patient and three surgeries in another patient) as well as different chemotherapy regimens. Two patients underwent OTL.
The 5-year OS rate in this cohort of patients is 57% (95% CI, 39–75%; Fig. 3) with a 5-year EFS rate of 28% (95% CI, 12–44%; Fig 4). The small size of the study population prevents the use of multivariate analyses to determine whether there were clinical characteristics (i.e., gender, patient age, α-FP level, and platelet values; different modalities of diagnosing lung metastases: CT scan positive only vs. CT scan and chest X-ray positive; PRETEXT categories; presence of single or of multiple tumor nodule(s) within the liver and numbers of lung metastases) that can predict different outcomes within this cohort of children. However, by plotting these characteristics in a “two-entrance table” (data not shown), it appears that, among the 7 patients who presented with just a single lung metastases, 6 are alive with NED, whereas only 9 have been censored as alive with NED among the 22 patients who presented with multiple and bilateral lung nodules (for 2 patients, no information is available on the actual number of lung nodules). Thus, it seems that a single lung nodule may predict a more favorable outcome than multiple and bilateral lung deposits. Finally, the 5-year OS rates of those patients who had a positive finding on only their CT scans with negative chest X-rays compared with the patients who had positive CT scans and chest X-rays, were 77% (95% CI, 54–91%) and 42% (95% CI, 19–66%), respectively; the corresponding figures for the 5-year EFS rate were 38% (95% CI, 12–65%) and 21% (95% CI, 1–40%). These differences, however, are not statistically different. Out of the seven children who presented with just a single lung nodule (see above), two had both a positive chest X-ray and a positive CT scan, and both are alive with NED; five patients had only a positive chest CT scan, and four of these five children are alive with NED, but only two of these children had PLADO alone. Both of the other two patients who were cured by PLADO followed by other salvage therapies and who are alive with NED had a histologically proven lung recurrence. The magnitude of the α-FP level decrease under chemotherapy also was investigated using the parameter proposed by Van Tornout et al.,10 but no conclusive findings were documented.
Therapy was well tolerated. No fatal toxic events were reported. A detailed review of the toxicity of the PLADO therapy is in preparation.
The therapeutic strategy adopted for the SIOPEL 1 study for those children with HB who presented with lung metastases was not very effective, with a 5-year EFS rate of only 28% (95% CI, 12–44%). This rate is significantly worse than that obtained in patients with localized HB (77%; 95% CI, 69–85%).6 In the SIOPEL 1 study, the presence of metastases resulted the most significant negative prognostic factor for the 5-year EFS rate.6 The overall positive response rate to PLADO obtained in these children and, particularly, the complete remission rate for lung metastases were actually quite encouraging, at 84% (95% CI, 69–92%) and 55% (95% CI, 52–58%), respectively; however, fewer than half of those children who achieved complete remission of metastatic disease were capable of maintaining it with protocol therapy alone. More precisely, a sustained remission from lung metastases was achieved in only 26% of the study cases. Thus, it appears that the principal reason for protocol therapy failure was the lack of control of the lung disease itself. Actually, considering that almost 70% of the children in this series who succumbed to their disease never had their lungs cleared, this also seems to be the main reason why the prognosis for these children at the present standard of care is so poor.
The unsatisfactory therapeutic results that we achieved with the protocol therapy in terms of EFS are shared uniformly by the other modern, CDDP-based, cooperative trials run in Europe and in the United States on patients with HB (Table 2).1–5 In terms of the 5-year OS rate, the result of the current series of 57% (95% CI, 39–75%) actually is quite satisfactory. The corresponding rate for the nonmetastastic patients was 82% (95% CI, 74–89%).6 The fact that the 5-year OS rate almost doubles the 5-year EFS rate is because 9 of 23 children who failed protocol therapy (39%) presently are long term survivors after having been saved with further chemotherapy and surgery: thoracic surgery for removal of lung metastases in 4 children (with multiple thoracotomies in 2 children and a single thoracotomy in the other 2 children) and OLT in 2 children. Three of the four patients whose therapy included thoracotomies at presentation had multiple and bilateral lung metastases. Aggressive thoracic surgery has been advocated as a potential curative option for those patients with HB that recurs in the lungs.11–13 In 1999, Black et al.11 conducted a literature survey of children with primary malignant tumors who underwent thoracotomies for excision of pulmonary metastases and observed that 12 of the 15 patients (including 5 personal observations) with a known follow-up were alive with NED at 9–84 months from the time of disease recurrence. Feusner et al.,12 reviewing the Children's Cancer Study Group (CCSG) 881 study in which 10 of 33 patients with Stage I HB were enrolled and who developed lung metastases, concluded that an aggressive surgical approach to recurrent disease, when isolated to the lungs, can offer a realistic chance of long term cure. In fact, all three of the long term survivors underwent lung surgery (two patients underwent multiple surgeries). The role of thoracic surgery in treating children with lung metastases also is supported by experience with other childhood solid tumors.14, 15
Table 2. Children with Hepatoblastoma who Presented with Lung Metastases: Treatment Results of the Principal Prospective, Cooperative Studies
No. of patients enrolled/patients with metastases
Treatment strategy (chemotherapy regimen)
Survival for patients with metastases (according to the current study)
CCSG: Children's Cancer Study Group; POG: Paediatric Oncology Group; CGP-HB89: Cooperative German Pediatric Oncology Group; SIOPEL 1: International Society of Paediatric Oncology Group, Liver Tumour Study 1; OS: overall survival; EFS: event free survival; DFS: disease free survival; 95% CI: 95% confidence interval.
Primary surgery (ifosfamide, cisplatin, and doxorubicin)
DFS (median follow-up, 64 months) by Stage IV, 29%
Primary chemotherapy (cisplatin and doxorubicin)
5-yr OS, 57% (95% CI, 39–75%)
5-yr EFS, 28% (95% CI, 12–44%)
The finding of a possible role of OLT in the treatment of children with HB who present with lung disease is provocative. In the current series, as described above, the 3 children who underwent OLT are long term survivors (follow-up from diagnosis, 41+ months, 61+ months, and 72+ months, respectively, and 34+ months, 46+ months, and 52+ months, respectively, from the time of OLT). In this regard, it is worthwhile remembering that, as a whole, 9 of 11 children who underwent OLT in the SIOPEL 1 study are alive with NED at a follow-up from diagnosis ranging from 36 months to 54 months.16 Other liver transplantation programs claim to have successfully treated HB patients who presented with lung metastases.17 OTL, however, should be recommended with extreme caution and only by experienced transplantation teams for this group of children who must be considered at high risk for tumor recurrence.
No prognostic factor analyses could be conducted reliably in this small cohort of patients. A trend toward a more favorable outcome has been noted for those children whose metastatic disease has been revealed only by chest CT scan compared with those who had a positive chest X-ray and CT scan. Also, patients who present with only a single lung nodule seem to have a better outcome than those who present with multiple lung deposits. Four of the five children who were considered to be affected by a metastatic HB, based on the finding of a single lung nodule visible in the chest CT scan only, are alive with NED; however, only two of them were cured by PLADO chemotherapy alone, and the other two, who were rescued with other therapies, had a histologically proven lung deposit. Thus, at this stage of our knowledge, we do believe that both chest X-rays and contrast-enhanced lung CT scans should be recommended for the accurate staging of HB in children.
In the late 1980s and early 1990s, when, in the light of the very promising data on the effect of CDDP on HB, a new generation of clinical trials was launched in both the United States and Europe, no treatment stratification by any patients' clinical characteristics, including the presence of lung metastases, was planned (except in the U.S. studies for those rare cases of pure fetal HB that were completely resected). Based on the data presented and the scattered data available in the literature on the prognosis of patients with HB who present with lung metastases, this uniformity in treatment among HB patients does not seem justifiable. More should be tried for these patients in an attempt to offer them better chances for a cure. The SIOP Childhood Liver Tumour Study group, in its present generation of studies, already has introduced the concept of the existence among the HB children of different risk categories for treatment failure according to which therapy is modulated.18 More precisely, children who are affected by metastatic HB (as well as those with PRETEXT IV tumors) are considered to be at “high risk” and presently are treated according to an articulated, multidisciplinary treatment program that also includes an experimental, intensive, multiagent chemotherapeutic regimen based on the rapid, alternating use of myelotoxic and nonmyelotoxic agents along with thoracic surgery in patients with persisting lung disease. While waiting for a better insight into the intimate biological characteristics of these tumors that, hopefully, will help in developing a more appropriate therapeutic strategy, it will be up to the present clinical research to determine whether this is the “more” that these patients seem to need and also whether, within this group of children, different prognostic subsets of patients exist.