Gene expression profiling spares early breast cancer patients from adjuvant therapy: derived and validated in two population-based cohorts

Table 4 Prognosis of the clusters identified in the training and validation sets

Cohort		Risk cluster				P value^a
		Total	Low	Medium	High
Training set
Stockholm	n	159	49	51	59
	5-year event^b	38 (24%)	2 (4.2%)	7 (14%)	29 (49%)	<10^-7
Treated with adjuvant therapy
Stockholm tamoxifen only	N ^c	104	38	33	33
	5-year event	22 (21%)	0 (0%)	6 (18%)	16 (48%)	<10^-6
Uppsala node-positive	n	76	14	27	35
	5-year event	33 (43%)	2 (14%)	10 (37%)	21 (60%)	0.002
Untreated with adjuvant therapy
Stockholm	n ^c	33	7	10	16
	5-year event	11	1 (14%)	0 (0%)	11 (69%)	0.002
Uppsala node-negative	n	135	51	55	29
	5-year event	17 (12%)	3 (5.8%)	7 (13%)	7 (24%)	0.02
Van't veer	n	78^d	19	33	25
	5-year event	33	2 (17%)	18 (53%)	13 (55%)	0.01

^aP value for trend test.
^bDistant metastases or death within 5 years.
^cSubset of the Stockholm cohort: 82 patients with tamoxifen 2–5 years, eight patients with tamoxifen <2 years, nine patients with tamoxifen in sequence with megesterolacetat, and five patients with tamoxifen and goserelin.
^dOne patient was removed because of missing data. The 5-year event was distant metastases within 5 years.

ISSN: 1465-542X