Andrew E Teschendorff*, Ali Naderi, Nuno L Barbosa-Morais, Sarah E Pinder, Ian O Ellis, Sam Aparicio, James D Brenton and Carlos Caldas*
Corresponding authors: Andrew E Teschendorff email@example.com - Carlos Caldas firstname.lastname@example.org
Genome Biology 2006, 7:R101 doi:10.1186/gb-2006-7-10-r101
(2006-11-17 10:42) Swiss Institute of Bioinformatics
The authors wrote in the discussion:
"Intriguingly though, the grade-derived signature was not validated in a large available
cohort , raising doubts as to its wider applicability."
The grade signature was not applied to Wang et al. cohort in the Sotirou 2006 paper
because histologic grade data were not publicly available , and therefore the cohort
can not be used to answer our main question. That is, figure 2 of Sotiriou et al.
can not be produced without histologic grade information.
The grade-index can be applied to this cohort, with hazard ratio 2.53 (1.62-3.94),
under the cutoff that gives the same ratio of "poor" and "good" prognostic groups
as that in figure 3b of this commented paper (Teschendorff et al.), in order to make
the hazard ratios comparable. The HR above is similar to the one reported in figure
3b (2.38 [1.38-3.69]). Note that we use all 209 ER+ patients, while figure 3b of Teschendorff
et al. only used 144 patients.
Co-author of Sotiriou et al 2006 J Natl. Cancer Inst.
BioMed Central Ltd unless otherwise stated. Part of Springer Science+Business Media.