Table 4.

Difference in change in secondary outcomes (HDL-cholesterol, total cholesterol and triglycerides) according to use of AIT in a nationwide cohort of postmenopausal women with early breast cancer, January 2009 to December 2020, Denmark

	Women	Pre-breast cancer-valuea	Post-breast cancer-value		Change in secondary outcome	Difference in change in secondary outcome
	n	Mean (mmol/L)	n	Mean (mmol/L)	Mean (mmol/L)	Crude (mmol/L)	Adjustedb (mmol/L)
Secondary outcomes
HDL-cholesterol
No use of AIT	2542	1.71	10,423	1.68	-0.02	0 ref	0 ref
Use of AIT	7917	1.69	33,074	1.64	-0.05	-0.02(-0.03,-0.01)	-0.01(-0.03,0.001)
Total cholesterol
No use of AIT	2541	5.32	10,565	5.19	-0.14	0 ref	0 ref
Use of AIT	7917	5.29	33,576	5.14	-0.16	-0.03(-0.06,0.01)	-0.04(-0.09,-0.003)
Triglycerides
No use of AIT	2531	1.48	10,487	1.58	0.10	0 ref	0 ref
Use of AIT	7893	1.53	33,325	1.65	0.13	0.03(0.004,0.05)	0.02(-0.01,0.05)

AIT, aromatase inhibitor treatment.

Only women with a measured pre-breast cancer level of HDL-cholesterol, total cholesterol or triglycerides, respectively, and a measured post-breast cancer level of HDL-cholesterol, total cholesterol or triglycerides, were included in the correspondent analysis.

‘Use of AIT’ is defined as allocated and dispensed treatment with AIT, and ‘No use of AIT’ is defined as no allocated and no dispensed treatment with AIT; both as recorded in the clinical database of the Danish Breast Cancer Group (see Table S3 for further definitions).

^aThe pre-breast cancer LDL-cholesterol level is predicted using a linear effects model for all women with at least one LDL-cholesterol measurement prior to breast cancer diagnosis (see Supplementary Methods for further details); the difference between the mean of the observed LDL-cholesterol measurement prior to breast cancer diagnosis and the predicted one was < 0.02 mmol/L.

^bThe model included, as fixed effects, use of AIT, demographic characteristics (age [2-year intervals], calendar period [1-year intervals], education, region of residence, disposable household income, and cohabitory status), tumor characteristics (laterality, size, histology and malignancy grade, lymph node involvement, human epidermal growth factor receptor 2-status), other anti-cancer treatments (type of surgery, chemotherapy, and radiotherapy), the interaction between laterality and radiotherapy and the time of lipid-measurement; and a random intercept, to account for the inter-individual variability.