Capítulo IV: Equilibrio entre fases de compuestos
3. Sistema pseudo-cuaternario Monoglicéridos + Diglicéridos +
Mammographic density is a promising tool, with great potential for breast cancer prevention. However, most research has so far focused on static measures of density, giving insight into population-based relationships. Density is a dynamic phenotype, so repeated measures of density may be more informative for predicting individual breast cancer risk and for developing personalised breast cancer prevention strategies. Assessing individual women’s repeated measures could help to reveal within-women relationships between density and other breast cancer risk factors, to help to understand the aetiology of breast cancer development and the interacting influences of different risk factors. It may also provide information on risk of breast cancer for individual women and hence be useful for personalised breast cancer risk estimation. Consideration of changes in density may also be useful for indicating a woman’s response to breast cancer treatment, such as endocrine therapy. If reductions in risk are mirrored by reductions in density, change in density could be used as a potential biomarker for decrease in risk as a result of the drug. It is therefore hypothesised that changes in density may be of greater use in breast cancer prevention than fixed density measures.
Several studies have previously looked into the benefit of using repeated measures of density for breast cancer risk and prevention. A review of key studies is outlined below, along with further research ideas arising from the studies that formed the rationale for this thesis.
1.1.5.1 Repeated measures of mammographic density and other breast cancer risk factors (body mass index)
An important breast cancer risk factor and confounder of density is BMI. BMI is a well- established risk factor for postmenopausal women (170, 172, 173), but weight gain across premenopausal years has also been linked to an increased risk of postmenopausal breast c ancer (173, 247). However, this can be reversible with short-term weight-loss through dietary (248) or surgical (249) means. For example, the Iowa Women’s Health Study showed a 25-40% decrease in postmenopausal breast cancer risk in women who sustained a 5% loss of body weight
44 compared with women who continued to gain weight at different periods of time between 18 years of age and menopause (250). However, the effects of short-term weight-loss on density are less well understood.
There have been few studies assessing the effect of short-term weight change on density, particularly over the premenopausal years when a loss in weight is most effective. A dietary intervention study by Boyd et al. assessed the effect of a two year low-fat, high-carbohydrate diet on density, and found that women on the weight-loss intervention saw a reduction in Cumulus TA (2.4% reduction), whereas the control group had increased TA (0.3% increase), and DA decreased in the intervention group more so than the control group (6.1% reduction vs. 2.1% reduction, respectively). The reduction in dense area was particularly apparent in women who transitioned from pre- to postmenopausal or who remained premenopausal during the study (251). Other studies exploring the effect of more drastic weight-loss after bariatric surgery on premenopausal dense tissue have not shown any consistent effect of the weight-loss intervention on dense tissue (252, 253).
If weight-loss-induced reductions in risk are shown to be mediated by density, a reduction in density could act as a possible biomarker for risk reduction as a result of weight-loss and lifestyle interventions. However, with only one known study published to have previously assessed dietary-based weight-loss on density in premenopausal women, more studies are required to assess this mediating pathway, and to test this possible risk reduction biomarker.
1.1.5.2 Repeated measures of mammographic density for breast cancer risk estimation
Several studies have made use of repeated measures of density to predict breast cancer risk in populations of women attending screening (254-261). These have mainly focused on change in density between two serial mammograms and its effect on breast cancer risk. For instance, in a case-control study of 85 breast cancer cases and 85 matched controls in the Women at Risk (WAR) Columbia University study, Work et al. reported that Cumulus-assessed density between two pre-diagnostic mammograms (median 4 years apart) decreased in time with controls (p=0.004), but not with cases (p=0.6) (259). This suggested that a lack of density reduction over time may be indicative of a future risk of breast cancer. Another study from the Breast Cancer Surveillance Consortium (BCSC) tested whether changes in density between current and previous mammograms (average 3 years apart) were associated with risk of breast cancer. This study involved a large cohort of over 300,000 women screened at various US registries, with around 2600 subsequent breast cancers diagnosed during follow-up. Here, Kerlikowske et al. found that within-women changes in BI-RADS categories were associated with risk in women with previous BI-RADS categories I, II and III, but not for women with
45 previous BI-RADS category IV (258), suggesting a potential residual effect of high density. However, these interpretations were limited by the small number of women in the most extreme categories (for instance, only 0.1% of controls and 0.2% of cases moved from BI-RADS IV to I). Furthermore, no adjustments could be made for BMI which may have introduced negative confounding to the density-risk association.
The null effect seen in Kerlikowske et al.’s study in women with initially high density, was also reported in a study by vans Gils et al. (257). Fully computerised methods were used to measure density change over a 10 year period in over 100 postmenopausal breast cancer patients and 400 matched controls. This study found that women who started the study with high density (>25%) which decreased over time, experienced the same risk as women who had prolonged levels of high density. However, similar to Kerlikowske et al., very few women moved between the extreme density categories (only 12 women had initial density >25% which reduced to <5% during the study). Another key finding suggested that women whose density decreased from moderate (5-25%) to low (<5%), had (non-significantly) higher risk than women who had consistently low density (OR=1.9, 95% CI, 0.6 to 6.1). Compared with the consistently low group, women with consistently moderate density had an OR of 5.7 (95% CI, 2.2 to 15.2), and women whose density increased from moderate to high had an OR of 6.9 (95% CI, 2.1 to 22.9).
However, not all studies show an effect of change in serial density measurements on breast cancer risk. Longitudinal studies by Maskarinec et al. and Vachon et al. showed that changes in Cumulus percent density did not differ between women with and without breast cancer (255, 256). Nonetheless, both studies were limited by their collection of BMI information. Maskarinec et al. reported that many of their mammograms did not have corresponding BMI measurements taken at the same time as mammography, and Vachon et al. also reported differences in the timings of BMI assessments, with 17% of women having BMI data extracted over a year after their mammogram. BMI is not a static measurement and may have changed between the time of mammography and BMI assessment, potentially affecting the results.
Whilst, changes between two measures of density may have an effect on breast cancer risk, little is known as to whether repeated measures of density add information to risk estimation beyond what’s already explained by a woman’s current density. Only one other known study has evaluated this by assessing the predictive ability of using two density measures. Kerlikowske et al., again using data from the BCSC, found that the BCSC 5-year risk model better discriminated between cases and controls with a two-measure density predictor than with a one- measure density predictor (AUCs 0.640 vs. 0.635, respectively) (262). However, no studies have evaluated the benefit of including more than two density measures; particularly an
46 unlimited number of mammograms taken at arbitrary points in time, as would be seen in a screening environment.
1.1.5.3 Repeated measures of mammographic density for breast cancer risk with endocrine therapy interventions
IBIS-I was the first trial to show that change in density could reflect the beneficial effect of tamoxifen in the primary prevention of breast cancer. A nested case-control study within the trial assessed 123 breast cancer cases and 942 controls to test whether density reduction on tamoxifen was associated with risk of developing breast cancer. Cuzick et al. found that women who had at least a 10% reduction in VAS density in the first 12-18 months after the start of tamoxifen had an approximately 63% reduction in breast cancer risk compared with women on placebo, whilst women who experienced <10% density reduction on tamoxifen had no difference in risk compared with women on placebo (19). This result suggested that density change could be used as an early biomarker to assess the efficacy of prophylactic tamoxifen in order to predict a woman’s response to treatment. With the help of this biomarker, healthcare practitioners may advise women who see at least a 10% density reduction after 12-18 months of treatment to continue with their 5 year course of chemoprevention, whereas those who see a more modest reduction or increase in density might not be responding to treatment and would perhaps benefit from alternatives such as lifestyle interventions or chemoprevention with other SERMs or AIs (19).
Other studies have since tested the biomarker in the adjuvant setting for breast cancer patients on endocrine therapy for treatment of the disease. Some studies have suggested that a reduction in density may be used as a biomarker for breast cancer recurrence on tamoxifen (263, 264) and AIs (264), and others have suggested its use for predicting a reduction in mortality for tamoxifen treatment (265, 266). However, there are currently no systematic reviews focussing on the evidence to suggest that mammographic density reduction in women receiving endocrine therapy is a biomarker for breast cancer outcomes such as reduction in risk, recurrence, mortality and incidence of contralateral breast cancer. A review of this sort is essential to determine the strength of certainty for this biomarker before it can be implemented into clinical practice.
There is also very little evidence for the mammographic density biomarker in women treated with AIs, and there are no known studies in women on preventive AI therapy. The IBIS-II trial showed that the AI, anastrozole, reduced the risk of ER+ breast cancer in high-risk postmenopausal women by 60% (208), and it is a good resource to test this biomarker for preventive anastrozole therapy.
47 Previous studies assessing the effect of AIs on density have reported only modest (and often underpowered) results (267-270). In the preventive setting, the NCIC CTG MAP.1 prevention trial of letrozole vs. placebo found that 12 and 24 month changes in Cumulus-assessed PDA were small and similar between arms (12 months: mean PDA change -1.74 on letrozole, -0.24 on placebo (adjusted p=0.61); 24 months: mean PDA change -0.01 on letrozole, -1.32 on placebo (adjusted p=0.61)) (268). Vachon et al. also found similar results in a study of over 100 postmenopausal women (adjusted mean PDA change -1.0% on letrozole vs. -0.3% on placebo (p=0.58)) (270). The NCIC CTG MAP.2 prevention trial found similar results for exemestane (mean 12 month Cumulus-assessed PDA change: 0.56 on exemestane and 0.58 on placebo (adjusted mean difference between arms p=0.96), mean 24 month PDA change: -0.17 on exemestane and -2.93 on placebo (adjusted mean difference between arms p=0.52)) (269). Studies in the adjuvant setting have shown similar results (267), but there has been some suggestion of a small effect of AIs on volumetric density with a larger sample size (271). However, there are currently no studies testing the effect of AIs on density in the preventive setting with a similarly sufficient sample size. The IBIS-II trial could be an important resource for testing the effect of preventive AIs on density with the potential to provide an adequately sized sample of women.