Purpose: To determine whether the information gained from staging laparotomy can be predicted by imaging and/or clinical factors in children with Hodgkin's disease.
Patients and methods: Between 1986 and 1991, 216 consecutive pediatric patients with Hodgkin's disease underwent laparotomy and were treated on two concurrent protocols in a multiinstitutional cooperative group. All patients had computed tomography (CT) of the chest, abdomen, and pelvis. Clinical factors studied included sedimentation rate, B symptoms, histology, number and location of involved sites, sex, mediastinal involvement, and age. Pretreatment CTs were centrally reviewed in 88 cases for the presence and size of both supradiaphragmatic and infradiaphragmatic lymph nodes, intrinsic spleen lesions, and splenic size. Models were generated that were predictive of any abdominal disease, splenic involvement, extensive splenic involvement, and upstaging at the laparotomy. False-positive and false-negative rates were calculated.
Results: For the end point of any abdominal disease, a model based on B symptoms, histology, sedimentation rate, and number and location of involved sites was highly significant (P < .0001). However, the success in predicting abdominal disease in an individual patient was limited: false-negative rate, 26%; false-positive rate, 32%. Highly significant models based on clinical factors and/or radiographic findings were also generated to predict splenic involvement, extensive splenic involvement, and upstaging with laparotomy, but they also had high false-positive and false-negative rates.
Conclusion: Laparotomy findings cannot be predicted accurately in the majority of patients based on knowledge of CT findings and clinical factors.