Summary of findings 3. Predictive biomarker.

Women receiving endocrine therapy or not receiving endocrine therapy (control group): effect of mammographic density reduction in endocrine therapy group versus effect of mammographic density reduction in control group
Patient or population: women receiving endocrine therapy (selective oestrogen receptor modulator or aromatase inhibitor) or not receiving endocrine therapy (control group) Setting: prevention of breast cancer or treatment of early‐stage breast cancer Intervention: effect of mammographic density reduction in endocrine therapy group Comparison: effect of mammographic density reduction in control group
Outcomes	Impact	№ of participants (studies)	Certainty of the evidence (GRADE)
Prevention: incidence of invasive breast cancer and DCIS ‐ tamoxifen	OR 0.53 (95% CI 0.21 to 1.32) for an interaction between per cent density reduction (≥ 10% or < 10%) and prophylactic tamoxifen (visual assessment in 5% intervals on mammograms 12 to 18 months apart; Cuzick 2011a)	123 cases/942 controls (1 observational study)	⊕⊕⊝⊝ LOW 1
CI: confidence interval; DCIS: ductal carcinoma in situ; OR: odds ratio
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: We are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: Our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: We have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

¹Certainty of evidence was initially graded moderate because data for this outcome were from a randomised controlled trial. We downgraded the certainty of the evidence from moderate to low (GRADE downgrading factor ‐ imprecision): confidence interval is wide and includes the null effect; power to detect an interaction is limited by the sample size and number of cases.