br Dose scaled to range between and br Whole br
Dose scaled to range between 0 and 1
Dose scaled to range between 0 and 1
Fig. 6 Correlations between mean radiation doses to the whole heart and different cardiac segments for women receiving tangential or anterior electron regimens for breast cancer in Sweden (1958-2001) or Denmark (1978-2000). To display correlations, all doses are scaled to range between 0 and 1. Each panel includes 105 points to show all estimated mean doses from 7 left-sided regimens reconstructed on 10 computed tomography scans and 7 right-sided regimens reconstructed on 5 computed tomography scans. Panels that show correlations between left-anterior descending coronary artery and right coronary artery segment doses and doses to the left ventricular (LV) segments usually supplied by the left anterior descending coronary artery (LV: apical, LV: septal, LV: anterior) and right coronary artery (LV: inferior, LV: septal) are highlighted.
a Berkson error.26 For studies in which the outcome measure of interest is a continuous variable, Berkson er-rors do not bias the slope of the estimated dose response. In the present case, where the outcome measure is binary, Berkson errors in estimated doses may result in small biases to estimated dose-response relationships.26 This is in Gefitinib with classic measurement errors, which can lead to considerable attenuation of derived dose-response relationships.
Doses to different segments were highly correlated, which would make investigating whether myocardial injury resulted from exposure of the myocardium causing disruption of the microvasculature or from exposure of an artery (eg, LADCA causing arterial occlusion and downstream myocardial ischaemia) difficult. Interest-ingly, RCA segment doses were not correlated with doses to the LV segments that are usually supplied by the RCA, which may provide an opportunity to differentiate be-tween myocardial injury caused by microvascular injury versus that caused by macrovascular injury. Differences between doses actually delivered to individual segments
and surrogate doses assigned based on the typical CT scan (owing to positional and anatomical uncertainties) will also be correlated. They may be positively correlated, for example, two adjacent segments may move into or out of the high dose region together such as the LV apex and distal LAD coronary artery in left-tangential radiotherapy (Fig. 2A). Or, they may be negatively correlated, for example, if patient position changed slightly during left direct electron IMC radiotherapy, the proximal LAD might move out of the field, and the proximal RCA move into the field (Fig. 2C).
In cardiac radiation dosimetry, high correlations be-tween the estimated segment doses limit the ability to derive meaningful quantitative associations between doses and injury to specific cardiac segments. This difficulty is compounded by the fact that differences be-tween the actual and estimated doses are also strongly correlated. Therefore, using these to derive quantitative dose-response relationships would be inappropriate. However, typical CT-scan doses consistently indicated whether a particular regimen typically gave a high,
and our segment dose rankings were consistent with those in other publications.13,27e29 There was also consistency in the locations of hotspots within the ventricular myocardium and main coronary arteries for each regimen (Figs. 4 and 5). Therefore, segment-specific doses may be used to rank segments by higher-versus-lower doses. Subsequently, these rankings may be related to the risks of cardiac segment injury in patients who received radi-ation therapy in the past to ascertain if segments that receive higher doses have a higher risk of injury. These findings may also be relevant to studies of women receiving contemporary radiation therapy.
Our study has several strengths. First, detailed infor-mation was collated on regimens from several sources, including individual radiation therapy charts for 470 women. Second, doses to coronary artery segments were estimated rather than whole coronary arteries. Arteries are long, thin structures that track in different directions around the heart. Segment doses may be more meaningful because generally only 1 or 2 segments of an artery receive a substantial dose, and other segments receive only a scattered dose. Third, we verified that the segments described were those referred to by the cardiologists when reporting the location of the cardiac injury, so that segment doses could be directly related to the location of the damage.