Multivariate analysis of Hodgkin's disease prognosis. Fitness and use of a directly predictive equation.

Abstract:

:The main purpose of this work is to give methodologic details and further verification regarding the predictive system recently elaborated for Hodgkin's disease (HD) patients, by which, under standard accuracy and conventional treatment, patient survival can be directly estimated (within a given confidence interval) when six of the best prognostic factors are suitable computed. The factors are, in decreasing order of importance, erythrocyte sedimentation rate (ESR), clinical stage, histologic subtype, age, serum albumin and sex. They were selected on the basis of a multivariate analysis applied to the exponential model for survival time distribution that proved to fit the original data from 586 HD patients accurately. A linear equation with these six variables was elaborated to calculate the estimated mean (or median) survival time or the estimated probability of surviving a given time. In the present work the validity of that estimate was successfully checked on a new external group of 261 patients, observed at the University of Bologna. Staging and treatment of these patients were similar to those in the Pavia series, the only exception being the combined radiochemoterapeutic treatment in patients presenting the nodular sclerosis histologic subtype and early clinical stage. A very good agreement was found between estimated survival (according to the Pavia data) and observed survival (in the Bologna series) for patients with any combination of the six prognostic factors, except those including nodular sclerosis and early stage. In these patients the observed survival exceeded that expected. These results demonstrate that (a) the choice of the six prognostic factors and the statistical weight allotted them were correct, and (b) the predictive system is sensitive to effective therapy modifications, which can be identified even in selected patient subsets of a general series. Thus, this prognostic equation can really be a powerful tool in clinical HD research. Explicative notes for the correct use of the predictive equation, together with an example, are reported in the Appendix.

journal_name

Haematologica

journal_title

Haematologica

authors

Gobbi PG,Gobbi PG,Mazza P,Zinzani PL

subject

Has Abstract

pub_date

1989-01-01 00:00:00

pages

29-38

issue

1

eissn

0390-6078

issn

1592-8721

journal_volume

74

pub_type

杂志文章