Patterns of lymph node sampling and the impact of lymph node density in favorable histology Wilms tumor: An analysis of the national cancer database.

INTRODUCTION: There is controversy about the role of lymph node (LN) sampling or dissection in the management of favorable histology (FH) Wilms tumor (WT), specifically how it performed and how it may impact survival. OBJECTIVE: The objective of this study was to analyze factors affecting LN sampling patterns and the impact of LN yield and density (number of positive LNs/LNs examined) on overall survival (OS) in patients with advanced-stage favorable histology Wilms tumor (FHWT). METHODS: The National Cancer Database (NCDB) was queried for patients with FHWT during 2004-2013. Demographic, clinical and OS data were abstracted for those who underwent surgical resection. Poisson regression was performed to analyze how factors influenced LN yield. Patients with positive LNs had LN density calculated and were further analyzed. RESULTS: A total of 2340 patients met criteria, with a median age at diagnosis of 3 years (range 0-78 years). The median number of LNs examined was three (range 0-87). Lymph node yield was affected by age, race, insurance, tumor size, laterality, advanced stage, LN positivity, and institutional volume. A total of 390 (16.6%) patients had LN-positive disease. Median LN density for these LN-positive patients was 0.38 (range 0.02-1) (Summary Figure). Estimated 5-year OS was significantly improved for those with LN density ≤0.38 vs. >0.38 (94% vs. 84.6%, P = 0.012). In this population, on multivariate analysis, age and LN density were significant predictors of OS. DISCUSSION: It is difficult to compile large numbers of cases in rare diseases like WT, and fortunately a large administrative database such as the NCDB can serve as a great resource. However, administrative data come with inherent limitations such as missing data and inability to account for a variety of factors that may influence LN yield and/or OS (specimen designation, pathologist experience, surgeon experience/volume, institutional Children's Oncology Group (COG) association, etc.). In this specific disease, the American Joint Committee on Cancer staging (captured by the NCDB) is different than the COG WT staging system that is used clinically, and the NCDB does not capture oncologic outcomes beyond OS. CONCLUSIONS: In a review of the NCDB, various factors associated with LN yield and observed LN density were identified to be significantly associated with OS in patients with LN-positive FHWT. This reinforces the need for adequate LN sampling at the time of WT surgery, to maximize surgical disease control. It was proposed that LN density as a metric may allow for improved risk-stratification, and possibly allow for therapeutic reduction in a sub-set of patients with low LN density.

Cholecystokinin receptor positivity in children with chronic acalculous gallbladder dysfunction: a pilot study to investigate the etiology of chronic acalculous gallbladder dysfunction.

BACKGROUND: The etiology of chronic acalculous gallbladder dysfunction (CAGD) is unknown. However, cholecystectomy is being performed as treatment, based on gallbladder (GB) ejection fraction studies. The aim of this study was to examine the pathology and immunohistology of GBs from children with CAGD. METHODS: Children with a diagnosis of CAGD were identified. Control patients had their GB removed for nonbiliary indications. Immunoperoxidase staining was performed using rabbit antihuman cholecystokinin receptor (CCK-R) antibody. The pathologist was blinded to the study and controls. RESULTS: Fifteen children were evaluated: 6 children with CAGD and 9 controls. All children with CAGD had abnormal cholecystokinin-stimulated nuclear imaging. Ejection fractions ranged from 8% to 30%. All patients reported resolution of symptoms on follow-up at 6 months. Histopathology of the GB was normal for both the controls and children with CAGD. Both control and CAGD GBs demonstrated positive staining for CCK-R in the vascular endothelium and smooth muscle. Mucosal epithelial staining was only observed in 5 of 6 of GBs of children with CAGD. In the sixth GB, the epithelium was too necrotic to assess. CONCLUSIONS: In this pilot study, expression of CCK-Rs in the epithelial cells is noted in children with CAGD compared with controls. The significance of this finding requires further investigation.

Wilms tumor: progress and considerations for the surgeon.

Wilms tumor (WT) or nephroblastoma is the most common tumor of renal origin found in children. It accounts for 6% of all pediatric tumors and is the second most frequent intrabdominal solid organ tumor found in children. Initial survival rates in the early part of the last century was only 30%, but now long-term survival in both North America and European trials is approaching 85% with many low-stage tumors significantly higher. Treatment is now progressing towards "risk-based management"- based not only on stage and histology but also incorporating genetic markers [Dome JS, Grundy PE, Perlman EJ, Ehrlich PF, et al. Protocols for the renal tumors study. Childrens Oncology Group. [www.childrensoncologygroup.org. 2007.]. Within the multidisciplinary treatment team the surgeon plays a critical role in the diagnosis, staging and the surgeon's technical skills and judgment directs therapy and impacts outcome. The next generation of treatment for children with WT will focus on identifying subsets of patients who can be defined by some criterion as having a different outcome than their similar stage peers and who therefore require a variation in management. These include children with WT that have unsatisfactory long-term survival (less then 75%), patients of good survival but high potential for late effects and a final challenge are those children with both a poor survival and a high potential for late effects. This article presents a review of the most recent treatment considerations for WT with a focus on the surgeon's role to ensure a good outcome.

Monitoring performance: longterm impact of trauma verification and review.

BACKGROUND: This study documents how the verification process at a Level I pediatric trauma center affected patient care through changes in care indicators (CIs) from predesignation through four postverification time frames. An important component of any verification program is its effectiveness, not only at the time of verification but during the time between "examinations." To date, few data exist describing the interval periods and the progression and maturation of a trauma program after initial verification. STUDY DESIGN: Forty-seven distinct CIs were monitored monthly through data generated from the trauma registry. Six distinct time periods were identified. PRE (January, June, October 1997), trauma care without monitoring; VER (November 1999 to September 2000), preparation for verification; and four postverification periods: P1 (January to June 2001), P2 (July to December 2001), P3 (January to June 2002), and P4 (July to September 2002). RESULTS: Between 1997 and 2002, trauma admissions increased from 200 per year to 313 per year. Mortality rate and Injury Severity Score distributions remained unaltered. Statistically significant (p < 0.05) quantitative and qualitative changes were observed in numbers (percent) of patients reaching clinical criteria. These included prehospital, emergency department, and hospital-based trauma competencies. Trauma patient evaluation (including radiology) and disposition out of the emergency department (<120 minutes) improved in each study section and remained high during the postverification time period. There was a strong pair-wise correlation (p < 0.005, Cronbach alpha 0.8) between CNS charting and acquisition of head CAT scans. Pediatric ICU duration of stay increased in both the (summer) P2 and P4 time periods. Prehospital and emergency department fluid monitoring remained unsatisfactory. CONCLUSIONS: Statistically significant changes in patient care indicators were noted to improve during the trauma center designation process, and other key deficiencies were identified and addressed. Maintaining these improvements requires constant monitoring or performance may revert below accepted levels.

Endotracheal intubations in rural pediatric trauma patients.

BACKGROUND/PURPOSE: Evidence from urban trauma centers questions the efficacy of pediatric field endotracheal intubations (ETIs). It is recognized that in the rural environment, discovery, transport delays, and a paucity of pediatric expertise contribute to higher pediatric trauma mortality rates compared with urban environments. The purpose of our study was to determine the effectiveness of field ETI in rural pediatric trauma patients. METHODS: ETI attempts (field, referring hospital, trauma center [TC]) in trauma patients less than 19 years old were included. Prehospital and TC charts, including demographics, injury mechanism, indication, location, person performing, number of attempts, Glascow Coma Scale (GCS), complications from ETI, and outcome, were assessed. RESULTS: Between 1991 and 2000, 105 of 2,907 patients met study criteria. Paramedics, trauma flight nurses (field ETIs), emergency physicians, surgeons, and anesthesiologists performed the ETI. One hundred fifty-five ETIs (1 to 6 per patient) were attempted in 105 children. Fifty-seven percent of the ETIs were attempted in the field, 22% in transferring hospital, and 21% at the TC. Successful intubation on first attempt was 67% (field), 69% (referring hospital), and 95% (TC). Subsequent ETI attempts had failure rates of 50% (field) and 0% (referring hospital, TC). Indication for ETI included fear of losing airway control (37%), closed head injury (36.1%), respiratory rate less than 10 or greater than 40 (11.2%), cardiopulmonary arrest (6.5%), respiratory arrest (4.6%), and airway obstruction 4.6%. Only 9.3% of children could not be oxygenated or ventilated by bag valve mask (BVM) before ETI. Twenty-three percent had complications directly related to ETI (eg, aspiration). The relative risk of an airway complication was 2.5x higher with more then one ETI attempt (P <.05). Four percent of the airway complications occurred in TC, 29% (transferring hospital) and 66% (field, P <.05), respectively. Airway complications and multiple ETIs were associated with transport delay, lower GCS, longer hospital stay, and lower discharge GCS (P <.001) but independent of injury severity score, sex, age, and survival (P >.05). CONCLUSIONS: Multiple ETI attempts are associated with significant complications and may offer limited advantage over BVM and possibly may affect outcome. Indications for field intubations may require review especially in rural pediatric trauma.

Understanding work-related injuries in children: a perspective in West Virginia using the state-managed workers' compensation system.

BACKGROUND/PURPOSE: Little data exist that defines the consequences of occupational injuries in children. Traditional assessment of work-related injury is coupled with disability payments based on salary, which give little insight into etiology and severity. The authors hypothesize that the risk and pattern of occupational injuries in young workers are different then adults. METHODS: Claims from 1996 through 2000 were analyzed from the West Virginia Bureau of Workers Compensation. To define the significance of an injury, child and adult groups were subdivided into injuries that required surgery (ie, serious injuries). Current Procedural Terminology (CPT) codes for anesthesia and surgical procedures were cross referenced with the claims to ensure group designation. Relative risks (RR) were used to compare groups. RESULTS: Between 1996 and 2000, 364,063 claims were submitted, 14,093 in workers < or =19 years of age. Two hundred seventy claims in children required surgery. Serious injuries in children occur more often in boys 2.2x mainly in the (16 to 24 hours) evening (48% v 23.13%; P <.05) and in July/August (26.5 v 18.4; P <.001). Falls were the main mechanism of injury. Proportionately fingers (1.70x) and hands (1.64x, 1.6 to 1.7) were injured in children. Lacerations (3.4x), fractures (1.4x), and amputations (3.75x) frequently resulted in general anesthetic procedures, and the RR of these injuries were increased versus adults. Service, manufacturing, construction, and agriculture were the main injury-related occupations in children. CONCLUSIONS: For any job category, injuries in children have unique features, tend to be more serious, and require a surgical intervention proportionately more frequently than adults.

Should helical CT scanning of the thoracic cavity replace the conventional chest x-ray as a primary assessment tool in pediatric trauma? An efficacy and cost analysis.

BACKGROUND/PURPOSE: Findings from studies in the trauma literature suggest that thoracic computed tomography (TCT) scanning should replace conventional radiographs as an initial imaging modality. Limited data exist on the clinical utility and cost of TCT scans in pediatric trauma. Our current practice is to obtain TCT scans in those children at risk for thoracic injures. The purpose of this study is to examine what additional information TCT provides, how frequently it results in a change in clinical management, and a cost/benefit analysis. METHODS: Children 18 years old and younger that had both a Chest x-ray (CXR) and TCT scan in their initial workup were included. Indications for TCT scan were (1) any sign of thoracic injury on CXR, (2) pathologic findings on physical examination of the chest, and (3) high impact force to chest wall. A child may have had one or more indications for a TCT scan. RESULTS: Between 1996 and 2000, 45 of 1,638 trauma patients met study criteria. Indications for TCT included thoracic injury on CXR (n = 27), findings on physical examination (n = 8) and high-impact force (n = 33). In 18 of the 45 (40%), injuries were detected with TCT imaging but not on CXR. These included contusions (n = 12), hemothorax (n = 6), pneumothorax (n = 5), widened mediastinum (n = 4), rib fractures (n = 2), diaphragmatic rupture (n = 1), and aortic injury (n = 1). In 8 patients (17.7%) TCT imaging resulted in a change in clinical management. These included insertion of a chest tube (n = 5) aortography (n = 2) and operation (n = 1). Age, sex, injury severity score, mechanism, and indication for TCT could not predict differences between TCT and CXR (P >.05). In our institution, the cost of a TCT is $200, and the patient charge is $906 ($94 per CXR). Based on our study data 200 TCTs would need to be done for each clinically significant change, increasing patient ($180,000) and hospital ($39,600) costs. CONCLUSIONS: Helical TCT is a highly sensitive imaging modality for the thoracic cavity; however, routine CXR still provides clinically valuable information for the initial trauma evaluation at minimal cost. TCT should be reserved for selected cases and not as a primary imaging tool.

Monitoring intraoperative effectiveness of caudal analgesia through skin temperature variation.

BACKGROUND/PURPOSE: A reliable noninvasive intraoperative marker of caudal analgesia effectiveness remains elusive. Caudal analgesia causes sympathetic inhibition resulting in vasodilatation, increased blood flow, and a resultant increase in temperature in the affected dermatomes. The authors hypothesize that this change in temperature between the affected and unaffected dermatomes is measurable and may represent a noninvasive method of monitoring effectiveness of caudal analgesia. METHODS: Children undergoing lower abdominal surgery participated in the caudal or noncaudal (control) analgesia arm of the study. After induction of general anesthesia, 0.25% bupivicaine (1 mL/kg) was infiltrated for a field block in control patients or a caudal block in the experimental group. Temperature was measured at the C4 and L2 dermatomes starting after induction and 5 minutes before the caudal or field block and every 5 minutes after. T(o) is defined as the difference between the C4 and L2. Delta T (DeltaT) is the temperature variation between T(o). A change in the DeltaT is defined by an increase in the L2 temperature. RESULTS: Forty-six families enrolled (36 experimental, 10 control). The DeltaT for controls was 0.2+/-0.09 degrees C (SEM). Each child in the experimental group had 2 temperature measurements before the caudal with an average DeltaT of 0.3+/-0.07 degrees C (SEM), thus, were internal controls. A marked increase in DeltaT at 5 minutes 0.5+/-0.06 degrees C (SEM) and at 10 minutes 0.6+/-0.07 degrees C (SEM; P <.05) was noted in the experimental group. CONCLUSION: A significant transient change in temperature is observed after caudal analgesia and maybe a noninvasive marker of effectiveness. Further study may clarify its clinical utility.

The need for a statewide pediatric trauma program.

Development and maintenance of an effective regional trauma care system mandates on-going assessment of those at risk, patterns of injury and types of resources available. It is known that a significantly higher injury and traumatic death rates exists for children in a rural environment, and there is also evidence to support improved outcomes for children treated at verified trauma centers. While many still rely on practice-based statistics, we postulated that population-based statistics are much more reflective as to what is actually happening and provide crucial information on how improvements can be achieved. To test this theory, we reviewed all pediatric traumatic deaths for children 18 years old and younger from Jan. 1, 1990 to Dec. 31, 1998 at the Jon Michael Moore Trauma Center at West Virginia University, a rural pediatric trauma center. We compared this data to trauma mortality within the center's 13 county primary region from the Office of Vital Statistics. Our research revealed that mortality from pediatric trauma is higher in rural environments than in urban environments, and that population-based statistics more accurately reflect the true impact of what is actually happening in any given region. Age appears to be an important factor in determining which children are transferred to the trauma center and this may represent a critical factor in outcome.

Lessons learned from a failed multi-institutional randomized controlled study.

BACKGROUND/PURPOSE: In 1996, the Surgical Sections of the Children's Cancer Group (CCG) and the Pediatric Oncology Group (POG) received National Cancer Institute funding to conduct a prospective, randomized, controlled, surgeon-directed study to evaluate the role of minimally invasive surgery (MIS) in children with cancer. Because of lack of patient accrual, the study was closed in 1998. The purpose of this study is to evaluate and describe those factors that impacted on study failure to ensure future successful clinical trials. METHODS: One hundred forty surgeons representing the surgical membership of CCG and POG as well as 111 institutions within CCG and POG were asked to complete a questionnaire about the failed clinical trial. The questionnaire focused on study objectives, organization, and institutional review board (IRB) submission. It also examined the surgeon's ability to perform the minimal access operation, the influence of the pediatric oncologist, and the existence of preconceived biases by surgeons, oncologists, and families. Statistical analysis was performed as appropriate. RESULTS: Eighty-six of 140 (62%) surgeons responded to the questionnaire. Only 23% of the potential protocols were submitted for IRB approval. Of responding surgeons, 39% were not actively performing MIS when the study opened. A surgeon's support of the study was directly related to when the surgeon received the protocols (P <.001) and whether the participating surgeon was actively participating in MIS (P <.016). The oncologist's knowledge and support of the study affected IRB submission and approval (P <.02) and was influenced by whether MIS was practiced at the institution (P <.05). The majority of responding surgeons believed the experimental question was relevant (P <.05). However, responding surgeons believed that a preconceived bias existed within both their local surgical and oncology communities favoring a particular surgical approach (P <.001), but this bias did not extend to the families (P >.05). CONCLUSION: The study failed because of lack of accrual for a variety of reasons: failure to submit to the institution's IRB, lack of surgical expertise with MIS procedures, and preconceived surgeon bias toward either an endoscopic or traditional open approach.

Outcome of perforated necrotizing enterocolitis in the very low-birth weight neonate may be independent of the type of surgical treatment.

Perforated necrotizing enterocolitis (NEC) in the low-birth weight infant is now one of the most common surgical problems encountered in contemporary neonatal intensive care units. However, morbidity and mortality from NEC remain high, and the optimal surgical management of these infants remains controversial. Currently few data exist comparing the factors influencing outcome in very low-birth weight infants with perforated NEC treated by either local drainage or exploration. We hypothesize that survival of very low-birth weight neonates with perforated NEC may be more dependent on clinical status than on treatment modality. We present our experience treating a large cohort of infants weighing less than 1000 g with perforated NEC. A retrospective cohort study describes our experience with perforated NEC in very low-birth weight infants in a Level III neonatal intensive care unit. Between January 1991 and May 1998 a total of 70 newbo infants weighing less than 1000 g were evaluated and managed for perforated NEC. Comorbid factors were identified and calculated for each infant. Primary treatment was either local drainage or laparotomy. Statistical analysis was performed by Student's t test and multiple logistic regression. A multiple logistic regression model examined factors (comorbidities, number of comorbidities, and mode intervention) influencing outcome. A Kaplan-Meier survival analysis comparing survival versus number of comorbidities was performed. Twenty-two infants with an average weight of 679 g were treated by local drainage. Forty-eight infants with an average weight of 756 g were treated with exploratory laparotomy. Infants treated by local drainage had a higher cumulative number of comorbid factors (5.2+/-0.50 vs 3.7+/-0.29; P < 0.05) than those managed by operative exploration. Fourteen infants (63%) initially undergoing local drainage for perforated NEC survived. Of the 48 infants 36 operated on survived (75%). No single factor or combination of any comorbid factors was predictive of outcome. The total number of comorbidities for each neonate did reach statistical significance (P < 0.05). A greater likelihood of death was associated with a higher number of comorbidities. Survival with four or fewer comorbidities was 84 per cent, whereas survival with greater than six comorbidities was 30 per cent. The mean number of comorbidities was greater for drainage than for surgery, and for the same number of comorbidities the probability of survival tended to be greater for those treated with drainage than for those undergoing surgery. Multiple logistic regression analysis identified the total number of comorbidities as affecting outcome rather than treatment choice. This suggests therefore that selection of therapeutic options for the patient requires evaluating all factors that may impact survival rather than applying a single treatment strategy for all patients.

Emergency medical care in rural America.

The delivery of high-quality emergency care in a rural setting requires a conceptual framework quite different from that required in urban and suburban environments, given that available resources are limited in the rural setting. The intermittent and episodic nature of seriously ill and injured patients who present to rural emergency departments makes it difficult to plan, staff, and equip in order to provide emergency medical care at the same level seen at higher volume urban or suburban institutions. The objective of this article is to describe the distinctive nature and widely unrecognized features of emergency medicine in rural and remote areas, with a focus on clinical, workforce, and economic issues. We present recommendations for a shift in thinking and a call to action on behalf of all emergency medicine professionals that are based on a realistic assessment of the current status of emergency medicine and that are needed to develop and sustain high-quality emergency medical care in rural America.

Correlation between parental perception and actual childhood patterns of bicycle helmet use and riding practices: implications for designing injury prevention strategies.

BACKGROUND/PURPOSE: Bicycle injuries account for 10% of all pediatric traumatic deaths. Bicycle helmets have proven to decrease morbidity and mortality, yet trauma data show low helmet use among injured children. However, owning a bicycle helmet does not universally result in a child wearing a helmet. Furthermore, we hypothesize that parental perception of their children's use of the bicycle helmet may not reflect accurately true utilization by their child. To investigate this hypothesis the authors examined both parents' and their children's reports of bicycle ownership, supervision, riding patterns, and helmet use. METHODS: A random sample of grade 5 and 6 students (ages 8 to 12) and their parents were surveyed about bicycle ownership, riding patterns, supervision, and helmet use. The children and their guardians responded independently to the questionnaire. Statistical analysis was performed using the chi(2) test when indicated. RESULTS: Eighty-eight of 102 children (86%) responded. This represented 56% girls and 44% boys aged 8 to 12 years. Sixty-nine of 90 (77%) of the parents returned the survey. Ninety-six percent of the children owned a bicycle. A total of 87.5% of children owned a bicycle helmet. Eighty percent of the time children ride their bicycles on the road or sidewalk, with less then 20% on marked trails or parks. Parents reported that their children wear a helmet 90% of the time. In contrast, children report no helmet use in up to 61% of riding instances (P <.05). Parents themselves do not wear a helmet in greater then 60% when riding, which is correlated by their children. Seventy-one percent of the children report that they ride unsupervised the majority of the time. CONCLUSIONS: Bicycle and bicycle helmet ownership is high among this study group. There is a significant possibility that children will ride unsupervised, in at-risk situations, without wearing a helmet. Parental perceptions about bicycle helmet use by their children may not accurately reflect true utilization. In this study group parents appear as poor role models for their children. Injury prevention strategies need to focus on children and adults to improve effectiveness.

Wilms tumor: progress to date and future considerations.

Wilms tumor is the most common tumor of renal origin found in children. In the last 50 years, remarkable progress has been made in the treatment and understanding of children with Wilms tumor. Through the development of multiagent chemotherapy and cooperative pediatric interdisciplinary groups conducting large randomized controlled clinical trials, survival has improved dramatically. In the next century it is expected that 80% of children with Wilms tumor will be long-term survivors. Therapy is progressing towards a risk-based management based not only on stage and histology but also incorporated genetic markers. This article reviews progress to date and possible future directions in the treatment of Wilms Tumor.

Improved long-term survival with preoperative chemotherapy for hepatoblastoma.

In 1986, The Hospital for Sick Children (Toronto, Canada) began to use a standard preoperative chemotherapeutic regimen for patients who had unresectable hepatoblastoma. In 1988, we extended this protocol to all children who had hepatoblastoma. Of 25 children who presented with hepatoblastoma, 22 were eligible for protocol therapy. After percutaneous biopsy, cycles of cisplatin (20 mg/m2/d for 5 days) and Adriamycin (25 mg/m2/d for 3 days) were administered every 3 weeks by continuous intravenous infusion. A CAT scan was performed after the third cycle. Surgery was undertaken if response indicated that complete resection was possible. If not, a further one to three cycles were given until response was adequate. Postoperatively, therapy was continued for a total of six cycles. Twenty of twenty-two (91%) tumors responded to chemotherapy. Over half required only three cycles. Twenty hepatic resections (6 segmentectomies, 10 lobectomies, 4 trisegmentectomies) were performed. Preoperative therapy significantly reduced the extent of resection calculated to be necessary for complete excision at an initial diagnosis of the primary tumor in all but one. In the two children with inadequate response, total hepatectomy and transplant was necessary for complete resection. No deaths or operative delays were attributed to chemotherapy toxicity. Nineteen of 22 children (87%) are alive with no evidence of disease, including both transplant patients. One death was caused by intraoperative bleeding and the other two were caused by metastatic lung disease at 22 and 26 months, respectively. Twelve children, eight with tumors that would have been unresectable before effective chemotherapy, have had follow-up for more than 5 years. This protocol of preoperative chemotherapy appears to be safe and effective for most children who have hepatoblastoma.

Preoperative chemotherapy in hepatoblastoma.

Since January 1986, 15 children with hepatoblastoma received three to six cycles of chemotherapy with cisplatin (20 mg/m2/day x 5 days) and doxorubicin hydrochloride (25 mg2/m2/day x 3 days) every 3 weeks before surgery. The extent of the tumor was defined by computerized tomography scan of abdomen and chest and diagnosis confirmed by percutaneous liver biopsy before therapy was started. Tumors in 10 children were considered unresectable at diagnosis because of pulmonary metastases, extensive bilobar involvement, or venous involvement. Volume of tumor reduction ranged from 35% to 95% and was independent of tumor histologic findings. After a minimum of three chemotherapy cycles, excision was undertaken when tumor volume decreased to a size at which hepatic resection was feasible and safe. Chemotherapy complications that were not serious did not delay surgery or result in tumor growth during treatment. Complete surgical excision was possible in 13 children, including 10 who had had unresectable tumors and five with pulmonary metastases. Only three resections of more than one liver lobe were required, and partial lobectomy was possible in four children. One operative death and three postoperative complications, one severe, occurred. Cyclic chemotherapy was restarted 4 to 6 weeks after surgery until a total of six courses were given. Twelve children (eight whose tumors were originally unresectable) completed treatment 3 to 56 (median, 21) months ago and have no evidence of disease. Two other children currently undergoing therapy may have residual disease. The results to date far surpass historic survival rates, which ranged about 25%.