This study evaluated effects of set-up errors that might occur in daily practice of tangential breast irradiation on OARs. The concave part of the heart was included in the tangential radiation field first. As deep set-up error increases, dose change steeply increases in a certain range. On the other hand, since lungs are originally planned with the concave part already included in the tangential field, a quantitative dose increase occurs according to deep set-up error per mm. Based on our calculations, it seems possible, and maybe even probably, that (at least) some cardiac effects known to be attributed to low doses of radiation may indeed reflect an inadvertent delivery of higher doses of radiation due to set-up errors.
For heart and its sub-segments, DIBH was insensitive to deterioration of mean dose for the same set-up error compared to FB. It was more beneficial given that the original planned dose in DIBH was smaller than FB for those OARs. Based on a previous study showing that the relative risk of acute major coronary events is 7.4% per Gy,(5) DIBH can reduce major coronary events by roughly 5% for the same 5 mm deep set-up compared to FB by reducing 0.7 Gy of mean heart dose. Although results with DIBH were evaluated to be somewhat worse for the ipsilateral lung than those with FB, the difference was manageable considering the threshold dose of pulmonary toxicity.
It is known that each radiotherapy facility in Korea accounts for 25% of the burden of breast radiotherapy.(12) Of course, it would be ideal if all set-up errors can be corrected and the planned dose can be presented. However, there is a necessity about an action level as long as there are practical limitations. When we only examined statistical changes in mean value of heart dose, not the risk of cardiac toxicity, DIBH was at an acceptable level when deep set-up error was up to 5 mm. However, the degree of error occurrence needed to be strictly observed with FB. When this pattern of deep set-up error within 2.5 mm and 5 mm consistently developed in the entire radiotherapy period, our study showed that approximately 23% and 49% of heart dose could increase, respectively, in comparison with the original plan. In modern series, median mean heart doses for left side breast cancer applying conventional (50 Gy in 25 fractions, DIBH 27.8%) and hypofraction (42.6 Gy in 16 fractions, DIBH 14.6%) schedule were 2.16 Gy and 1.47 Gy, respectively.(13) When our study results of DIBH are applied to the above study, cardiac dose can rise up to 3.23 Gy and 2.20 Gy under condition of 5 mm deep set-up error, respectively. Automated heart edge detection in cine MV image has been proposed.(14) If such technology is commercialized, adaptive radiotherapy could be applied to systematically monitor cardiac dose so that cardiac dose can be controlled below the constraint of each institution.
It is expected that set-up error can be controlled within about 4 mm by utilizing currently developed technology. In comparison with conventional laser-based set-up, surface guided radiotherapy using optical surface scanning system (OSS) can significantly reduce set-up errors, showing that 95% of fractions are within the clinical action level of ≤ 4 mm in any direction.(15) Patients with frequent set-up errors require more thorough management. Patients with uncertainty of initial treatment associated with inter-fractional variation should be carefully observed in the entire treatment period.(16)
As DIBH requires holding the breath for more than 20 seconds and maintaining the same posture, reproducibility during radiotherapy is an important issue. A study estimating intra-fractional error using real time monitoring of OSS has presented that the mean motion during DIBH is small with < 1 mm translational and 1° rotational deviation.(17) In another study,(18) set-up error during DIBH was measured using continuous portal imaging in 58 patients. The group mean of fractional and intrafractional set-up error was within 1.7 mm of both systematic and random errors.(18) However, occasional large error exceeding 5 mm was occasionally presented in 2.2% of treatments.(18) Cardiac motion affects cardiac dose. Distance variation from systolic to diastole was ≤ 4 mm for the LV and ≤ 3 mm for the heart and the LAD with a maximum dose of 5.2 Gy for the LV and a mean dose difference of 4.6 Gy for the LAD.(19)
For lung cancer, lung dose constraint such as mean lung dose < 20 Gy, V20 Gy < 30%, V5 Gy < 65%, and absolute volume lung spared > 5 Gy, < 500 ml was recommended to protect radiation induced lung injury.(20) In a systemic review of recent reports regarding lung dose of breast radiotherapy, the average mean ipsilateral lung dose was 8.4 Gy for whole breast radiotherapy without breathing adaptation.(21) Therefore, the occurrence of lung toxicity is modest. Symptomatic pulmonary events of grade 2 developed in 2.7% of whole breast radiotherapy in actual modern practice.(22) The increase of mean lung dose was significantly correlated with lower lung volume and larger treatment volume.(23) In our study, the mean ipsilateral lung dose increased approximately 1 Gy per 2.5 mm deeper set-up, reaching 10 Gy provided that the action level was 5 mm deep for the heart.
This study considered the coverage of CTV as the most important factor when making a radiation plan without modifying the plan according to the proximity of the heart. Due to such principle, the original plan dose was somewhat high, especially for LAD. The DEGRO expert panel recommends cardiac dose constraints as mean heart dose < 2.5 Gy; mean LV dose < 3 Gy; V5 Gy of LV < 17%; V23 Gy of LV < 5%; mean LAD dose < 10 Gy; V30 Gy of LAD < 2%; and V40 Gy of LAD < 1%.(24) In actual treatment, if a radiation field is tailored by weighting the location of tumor bed and heart toxicity, it will be possible to maintain a cardiac dose as low as reasonably achievable. In addition, since this study was not a comparative evaluation of the set-up of DIBH and FB in the same patient, there might be errors depending on body contour of the selected patient. Lastly, it is important to note that in actual treatment, uncertainty of set-up may complexly occur besides our deep and caudal set-up. However, the evaluation was performed on the premise of a deep and caudal set-up.