Table 1. A summary of the AI trials that have been reported to date.

	n	AI	Design	Age	ER+	% n+	FU	DFS HR	Events AI/TAM	DFS P val	OS	Deaths AI/TAM	OS P val	ref
ATAC	5216	A	sub	64	100	35	100	0.85	618/702	0.003	0.97	472/477	NS	1
BIG 1-98	4922	L	sub	61	98	41	76	0.88	509/565	0.03	0.87	303/343	0.08	2
IES	4602	E	sw	n/s	100	48	56	0.75	339/439	0.0001	0.83	210/251	0.05	3
ABCSG-8	2922	A	sw	n/s	100	25	72	0.79	202/235	0.038	0.77	130/157	0.025	4
ARNO-95	979	A	sw	61	97	25	30	0.66	36/47	0.049	0.53	15/28	0.045	5
ITA/GROCTA	828	A,AG	sw	64	100	85	78		not stated		0.61	48/74	0.007	6
NSABP B33	1598	E	ext	60	100	48	30	0.68	37/52	0.07	1.0	16/13	NS	7
MA17	5187	L	ext	62	97	46	64	0.68	164/235	0.0001	1.0	154/155	NS	8

Abbreviations: sub=substitution, sw=switching, ext=extension. A=Anastrozole, L=Letrozole, E=Exemestane, AG=Aminogluthemide.

Those trials that have a statistically significant mortality benefit are highlighted in pink. This table clearly shows that only the switching trials have been able to show a mortality benefit from the AIs.

Refs: 1=ATAC Trialists, 2008; 2=Mouridsen et al, 2008; 3=Coombes et al, 2007; 4=Jakesz et al, 2008; 5=Kaufmann et al, 2007; 6=Boccardo et al, 2007; 7=Mamounas et al, 2008; 8=Ingle et al, 2008a.

All these analyses are intent to treat (ITT) with no adjustment for crossover and patients are kept in their originally assigned groups even if they crossed over from the control arm to the investigational arm. However, for the ATAC and IES data presented in this table, the ER unknown or ER-negative patients are excluded from analysis.

For MA17, a 2005 report showed an OS advantage in the node-positive subgroup; this was not seen in the most recent results, probably because of a high crossover rate.

For ATAC, the figure in the table is for the ER-positive subgroup.