Background: Mammographic breast density and parenchymal patterns are well-established risk factors for breast\ncancer. We aimed to report inter-observer agreement on three different subjective ways of assessing mammographic\ndensity and parenchymal pattern, and secondarily to examine what potential impact reproducibility has on relative risk\nestimates of breast cancer.\nMethods: This retrospective caseââ?¬â??control study included 122 cases and 262 age- and time matched controls (765\nbreasts) based on a 2007 screening cohort of 14,736 women with negative screening mammograms from Bispebjerg\nHospital, Copenhagen. Digitised randomized film-based mammograms were classified independently by two readers\naccording to two radiological visual classifications (BI-RADS and TabÃ?¡r) and a computerized interactive threshold\ntechnique measuring area-based percent mammographic density (denoted PMD). Kappa statistics, Intraclass\nCorrelation Coefficient (ICC) (equivalent to weighted kappa), Pearsonââ?¬â?¢s linear correlation coefficient and limits-ofagreement\nanalysis were used to evaluate inter-observer agreement. High/low-risk agreement was also determined by\ndefining the following categories as high-risk: BI-RADSââ?¬â?¢s D3 and D4, TabÃ?¡rââ?¬â?¢s PIV and PV and the upper two quartiles\n(within density range) of PMD. The relative risk of breast cancer was estimated using logistic regression to calculate\nodds ratios (ORs) adjusted for age, which were compared between the two readers.\nResults: Substantial inter-observer agreement was seen for BI-RADS and TabÃ?¡r (?=0.68 and 0.64) and agreement was\nalmost perfect when ICC was calculated for the ordinal BI-RADS scale (ICC=0.88) and the continuous PMD measure\n(ICC=0.93). The two readers judged 5% (PMD), 10% (TabÃ?¡r) and 13% (BI-RADS) of the women to different high/low-risk\ncategories, respectively. Inter-reader variability showed different impact on the relative risk of breast cancer estimated\nby the two readers on a multiple-category scale, however, not on a high/low-risk scale. TabÃ?¡rââ?¬â?¢s pattern IV demonstrated\nthe highest ORs of all density patterns investigated.\nConclusions: Our study shows the TabÃ?¡r classification has comparable inter-observer reproducibility with well tested\ndensity methods, and confirms the association between TabÃ?¡rââ?¬â?¢s PIV and breast cancer. In spite of comparable high\ninter-observer agreement for all three methods, impact on ORs for breast cancer seems to differ according to the\ndensity scale used. Automated computerized techniques are needed to fully overcome the impact of subjectivity
Loading....