Dummy run quality assurance study in the Korean Radiation Oncology Group 19 − 09 multi-institutional prospective cohort study of breast cancer

In this dummy run study, we found variations in radiation dose delivered to target volumes and organs at risk between institutions. As KROG 19 − 09 is a prospective cohort study, we accepted the dosimetric variation among the different institutions. Although IMRT is available at all the participating institutions, a majority use 3D plans for breast cancer because of limitations in resources. Some institutions still use anatomical landmarks without CTV contouring to reduce workload. The idea that irradiation of the entire lymphatic system may not be necessary for oncologic benefit and the low incidence of severe toxicity with breast RT plans are cited as reasons for using anatomical landmarks without CTV contouring. With a 3D plan, it is inevitable that unintentional doses are delivered to unintended areas. Another limitation of 3D plans is an unavoidable low dose at the field junction. Differences in RT techniques are the main reason for the dose variations between institutions.

As shown in previous international trial dummy runs, it is essential to implement quality assurance to allow the quality of trial data to be optimized and quantified [12,13,14]. In 2017, a phase III randomized trial was initiated by the Korean Radiation Oncology Group (the KROG 17 − 01 study, NCT03269981) to analyze the impact of RNI in pN1 breast cancer patients receiving effective systemic therapy. The primary objective of the KROG 17 − 01 study was to compare disease-free survival between WBI and WBI + RNI in pN1 breast cancer patients who received BCS and taxane-based chemotherapy. For adequate interpretation of the KROG 17 − 01 study results, an in-silico planning study comparing radiation dose distributions to the regional lymph nodes between the WBI and WBI + RNI plans of institutions participating in the KROG 17 − 01 study was performed [15]. The study found that the relative nodal dose was significantly lower with WBI than WBI + RNI (p-value < 0.01) in all nodal regions. It also found moderate-to-strong agreement in radiotherapy treatment volumes between the participants. Significant proportions of radiation were unintentionally delivered to the axillary lymph node level 1 and IMN regions in the WBI plans. Our findings agree with the KROG 17 − 01 in-silico study.

Another dummy run study for quality assurance of a randomized trial on IMN irradiation (the KROG 08 − 06 study) reported that the mean radiation dose to the IMN region was 40–74% of the prescribed dose in their WBI arm [16]. In our study, the median D95% (%) of CTVn-IMN for the large- and medium-breast cases were 4.1% (range, 1.6-17.3%) and 3.6% (range, 1.4-15.5%), respectively, in the WBI plans. Compared to the KROG 08 − 06 dummy run study, the radiation dose to the IMN region in the WBI alone plan was lower in our study.

Recently an insightful dose evaluation study was published [5]. The study reconstructed the treatment plans of the landmark Z0011 [17], AMAROS [18], EORTC 22,922 − 10,925 [19], and MA.20 [20] randomized lymph node irradiation trials to assess the dose distribution to actual lymph node metastases and the ESTRO-CTVs. They searched the study protocols of the Z0011, AMAROS, EORTC, and MA.20 trials for specifications regarding the treatment planning procedure. The field arrangements described in the protocols were used to imitate the 2D treatment plans on 3D computed tomography datasets of (1) a standard patient, (2) an obese patient with large breasts, and (3) a slender patient with small breasts. In these landmark trials, dose distributions at the axillary level 1, 2, 3 and the supraclavicular and IMN regions varied. These variations resulted from differences in RT techniques and field designs in each trial. They also found that the extent of incidental irradiation to the axillary nodes depended clearly on the patient’s body shape. In line with this previous study, we found inter-institutional and inter-case dose variations.

Variations in dose distribution at the regional nodal areas and OARs between participating institutions may affect the KROG 19 − 09 study’s clinical outcomes. To reduce the dosimetric variation among institutions, we plan to provide feedback from periodic audits. Furthermore, actual patient RT plan data should be collected to analyze the effect of these variations on regional recurrence rates and toxicity. To ensure reliable results, participants of KROG 19 − 09 agreed to the collection of actual patient RT plan data. We plan to collect DVH data by using artificial intelligence (AI)-based auto contouring software to delineate CTV and OAR structures. We will be able to analyze the relationship between the radiation dose and recurrence or toxicity rates based on the segmented structures with these data.

There are several limitations to this study. First, we performed the dummy run study in only two different breast-sized cases because of resource limitations. The inclusion of more cases with more diverse breast sizes could have provided more information about dose variation. Second, some institutions dropped out of and others joined KROG 19 − 09 after this dummy run study. Therefore, the results of this dummy run study are not able to reflect all the participants of KROG 19 − 09. However, this study is worthwhile as it allowed the participants to agree on collecting actual patient RT planning data in order to obtain reliable results for KROG 19 − 09.

In conclusion, in this dummy run study, we found inter-institutional and inter-case variations in radiation dose delivered to target volumes and organs at risk. As KROG 19 − 09 is a prospective cohort study, we accepted the dosimetric variation among the different institutions. Actual patient RT plan data should be collected to achieve reliable KROG 19 − 09 study results.

留言 (0)

沒有登入
gif