Use of Bootstrapping to Evaluate Infrequent Genetic Abnormalities as Prognostic Factors for Survival in Human Acute Leukemias
The distributional properties of a summary random variable in a two-sample test involving a small group can be investigated through bootstrapping. With data from a prospective Cancer and Leukemia Group B (CALGB 8461) study of the prognostic significance of cytogenetic abnormalities on human acute leukemias, survival data involving an infrequent genetic abnormality was bootstrapped using Cox’s proportional hazards regression model. Eight of 578 cases in the survival analysis exhibited trisomy 13 as the sole cytogenetic abnormality. Although the bootstrap results agree with the classical log-rank test, the distribution of bootstrap regression coefficients deviates from normality (P< 0.01, Kolmogorov-Smirnov). This bootstrap methodology can be of value for exploratory studies of survival involving a small group to assess the reasonableness of the log-rank statistic when the asymptotic assumption is questionable.
Unable to display preview. Download preview PDF.
- Chang I, Pagano M (1980), ”An Algorithm for Maximizing Cox’s Likelihood Function for Censored Data,” Technical Report No. 562, Sidney Färber Cancer Institute, Division of Biostatistics, Boston.Google Scholar
- Döhner H, Arthur D, Ball E, Sobol R, Davey F, Lawrence D, Gordon L, Patii S, Wurana R, Testa J, Verma R, Schiffer C, Wurster-Hill D, Bloomfield C (1990), ”Trisomy 13: A New Recurring Chromosome Abnormality in Acute Leukemias,” Blood 76, 1614–1621.Google Scholar
- Efron B (1981), ”The Jackknife, the Bootstrap and Other Resampling Plans,” SIAM, Philadelphia.Google Scholar
- Efron B, Tibshirani R (1985), ”The Bootstrap Method for Assessing Statistical Accuracy,” Technical Report No. 101, Stanford University, Division of Biostatistics, Stanford.Google Scholar
- Emrich L, Reese P, Kalbfleisch J (1981), ”Cox Model: A Proportional Hazards Model Analysis Package for SPSS Users,” Proceedings of the Fifth Annual SPSS Users and Coordinators Conference, 215–237.Google Scholar
- Noreen E (1989), ”Computer Intensive Methods for Testing Hypotheses,” J. Wiley, New York.Google Scholar