Data on overall survival (OS) and transformation to diffuse large B-cell lymphoma were obtained from 2 distinct sources. Ninety-six of 548 patients were observed to transform; however, 6 of these patients transformed after the last OS follow-up time. This was not noticed during analysis; since it is possible that including these 6 in time-to-transformation analysis may render the censoring mechanism nonrandom, the authors have repeated this analysis, censoring these 6 patients at the end of their OS follow-up time.
As a result, on pages 5722-5724, under “FL heterogeneity and transformation,” the P value for the cluster that was not predictive of time to transformation should be .8, not .9. Also, for gene TP53, the hazard ratio (HR) should be 2.95, not 3.13, and its 95% confidence interval (95% CI) should be 1.55-5.50, not 1.71-5.76. For CDKN2A, the HR should be 2.65, not 2.50, and its 95% CI should be 1.22-5.74, not 1.16-5.41. For BTK, the HR should be 2.17, not 2.07, its 95% CI should be 1.00-4.69, not 0.96-4.48, and its P-value should be .05, not .064. For GNA13, the HR should be 1.48, not 1.76, its 95% CI should be 0.74-2.94, not 0.94-3.29, and its P-value should be .27, not .080. For IRF8, the HR should be 0.52, not 0.50, its 95% CI should be 0.24-1.13, not 0.23-1.07, and its P-value should be 0.10, not 0.74. TP53, CDKN2A and BTK retain their reported associations, but the authors withdraw any claim about GNA13 or IRF8. Finally, the P value for the evidence for the association between time to transformation and nonaberrant somatic hypermutation burden should be 0.038, not 0.052.
In addition, the authors have prepared a new version of the supplemental Tables, removing censoring information (derived from when the transformation data were uploaded) from the transformation data. Since transformation is reported only for patients who transformed, no inference (such as whether they are still alive) can be made about subjects not reported as having transformed, other than that they have not transformed. Any analysis of time to transformation should censor the transformation times at the last observed survival time.
The corrected supplemental Tables are available in the HTML version of this erratum.
