Exploring cancer register data to find risk factors for recurrence of breast cancer – application of Canonical Correlation Analysis

Table 3 Canonical Structure Matrix for Predictor and outcome Variates

Predictor Set	U₁	Outcome Set^‡	V₁
Age	.223	DM, first two years	.837
Tumor location		DM, 2–4 years	.332
Superior medial	.138	DM, more than 4 years	.193
Inferior medial	.159	LRR, first two years	.486
Superior lateral	-.056	LRR, 2–4 years	-.030
Inferior lateral	.155	LRR, more than 4 years	-.013
Nipple area	.160
Side	-.017
Tumor size *	.432
LN involvement *	.567
LN involvement (N0) ^†	.580
NHG	.697
Perigland growth *	.566
Multiple tumors *	.110
Estrogen receptor	.370
Progesterone receptor	.365
S-phase fraction	.629
DNA index	.325
DNA ploidy	.342

Abbreviations: LN: lymph node, NHG: Nottingham Histologic Grade, DM: Distant Metastasis, LRR: Loco-regional Recurrence
* from pathology report, ^† N0: Not palpable LN metastasis, ^‡ all periods are time after diagnosis.
If the signs in the sets are the same, then if one increases the other also increases, and vice versa.

ISSN: 1472-6947