There is a great need to identify patients at risk of RIHD after radiotherapy in order to institute preventive measures and early interventions. It was with the goal of determining the usefulness of CMRs in the detection of preclinical RIHD that the current pilot study was undertaken. CMR appeared to be a promising candidate towards a non-invasive screening test in patients at high risk for RIHD.[21–26, 28] Despite long intervals and high radiation doses in several patients in our study, no association between radiation dose and myocardial changes was identified using these techniques.
The current study did not demonstrate significant correlations between LGE and radiation doses. We did not observe a dose-volume relationship between previously delivered dose distributions and areas of LGE. Included in our analysis was a wide range of known predictive factors for RIHD, including whole heart and left ventricle V5Gy, V30Gy, V40Gy, and V50Gy according to data from Bradley et al., Chun et al., Bogaard et al., and Speirs et al.[13, 15, 36, 37] All were found to not be significantly correlated with LGE presence. The location of cardiac Dmax was located outside the LGE volume in all patients with demonstrated LGE. There was also no association between cardiac Dmax and Dmean and the LGE volume. Based on this study, myocardial fibrosis and tissue remodeling following radiation treatment may not occur in high dose regions of radiation treatment plans.
In addition, diffuse T1 mapping did not show a clear dose-response relationship despite high cardiac radiation doses and multi-year follow up in some patients. There was a trend towards higher T1 values in patients with evidence of LGE, which likely shows the effect of local fibrosis changes on mean T1 values obtained from larger volumes.
On our review, the current study is one of only few to evaluate LGE and T1 mapping sequences in the detection of diffuse left ventricular fibrosis secondary to radiotherapy. Our analysis on LGE most similarly parallels the recently published manuscript by Huang et al.[38] In Huang et al., 7 patients with previous thoracic radiotherapy with median time from treatment to CMR of 3.1 years exhibited a linear relationship between EQD2 Dmean and Dmax doses delivered to the left atrium and right atrium and the fibrosis volume on CMR. Interestingly, however, they did not observe any focal left or right ventricle myocardial fibrosis, and did not see a dose relationship within the right atrium scar volume.[38] In our LGE protocol, no images of the right or left atria were acquired as atrial walls are generally thin and reliable LGE measurement is therefore difficult. Interestingly, despite higher prescription doses in our study no dose effect was identified for LGE.
In a study by Umezawa et al [39], 24 esophageal cancer patients who were treated with concurrent radio-chemotherapy to 66–70 Gy received CMRs at a median time of 23.5 months after completion of radiotherapy. All patients had 3D conformal treatments. Fifty percent of those patients demonstrated LGE, with LGE being localized always within the segments within either the predominantly 40 Gy (15.38% with positive LGE of all 40 Gy segments) or 60 Gy (21.21% with positive LGE of all 60 Gy segments) isodose distributions. In a follow up study by the same group, Takagi et al. prospectively enrolled 14 patients with newly diagnosed esophageal cancer, with serial CMRs taken before, 0.5 years and 1.5 years after 50.4–57.4 Gy chemoradiotherapy treated with 3D conformal techniques. LGE was detected in one baseline pretreatment CMR, with 78% (11/14) demonstrating LGE 1.5 years after chemoradiotherapy, mostly detected in the basal septum corresponding to high dose regions. While cardiac radiation doses were not stated, given 3D conformal treatments and esophageal cancer primaries, patients on the Umezawa et al. and Takagi et al. studies likely had higher mean cardiac doses than our cohort, as well as larger and more homogeneously irradiated cardiac volumes. In comparison, about half of the patients on our study had either IMRT or SBRT treatments with inhomogeneous dose distributions and small Dmax volumes, with the remaining 3D treatments primarily being tangential fields for breast radiotherapy with minimal cardiac overlap and little dose to the septum or right heart (Table 1).
Even less information is available on T1 mapping and RIHD. A case study describes a 70 year old male treated with chemoradiotherapy for esophageal cancer, who developed LV systolic failure 8 years after cancer treatment. CMR detected a T1 time of 1303 ms, which was the upper limit of reference range per the authors’ institution. Confirmatory endocardial biopsy showed interstitial fibrosis and myocardial degeneration compatible with RIHD.[40] The significance of this T1 value is unclear, as there is a lack of standardization and a true reference range is currently not established. Individual T1 values can not be compared between studies due to differences between vendors, CMR sequences, and post-processing.[26, 41] Tuohenin et al. studied 20 patients with early stage left-sided breast cancer received CMR with T1 mapping 6 years after radiotherapy. Diffuse T1 relaxation times were on average 1210 ms (+/- 52 ms) within inferoseptal segments of left ventricles which corresponded to radiation treatment fields, 35% of T1 values in this region were greater than 1250 ms, significantly greater than in other regions.[42]
The most comprehensive study on T1 mapping in the detection of RIHD is also by Takagi et al as described above. Mean T1 values for CMR at baseline prior to radiotherapy were 1183 ms (+/- 46); however, values taken at the basal septum which received sizeable radiation doses were significantly elevated above baseline pretreatment values (0.5 years = 1257 ms (p < 0.01), 1.5 years = 1238 ms (p = 0.024)). Interestingly, the left lateral ventricle segment, which would have been outside of the traditional 3D conformal field and received only low dose, did not see differences between pre-treatment and post-treatment T1 values.[43]
The time to the development of cardiac fibrosis is not known. The median time from end of treatment to CMR acquisition in our sample was 24.6 months, similar to the studies by Huang et al. and Umezawa et al. who detected fibrosis at a median 37.2 months and 23.5 months following radiotherapy, respectively. Patients demonstrating enhancement had a median interval between radiotherapy and CMR of 11.9 months compared to 31.4 months in patients without enhancement (p = 0.33), but the significance of this absolute difference is unclear and likely subject to bias and confounders.
There are several limitations to the current study. While RIHD can have many causes, CMR in this study primarily assessed myocardial damage and therefore cardiomyopathy. Due to the cross-sectional study design, no pre-radiotherapy CMRs were acquired and therefore areas of LGE on post-treatment CMRs may actually be related to “silent” undiagnosed cardiac events or may have been pre-existing. In addition, we cannot exclude that chemotherapy might have been related to LGE development. Although not demonstrated in the present study, with radiotherapy being a localized treatment a correlation between the location and volume of the LGE with the high dose distribution was anticipated whereas for systemic treatments more diffuse myocardial effects would be expected. While ideally investigating one primary tumor site only would have provided a more homogeneous patient cohort, dose standardization and investigation of a large spectrum of dose parameters in the present study are expected to account for interpatient differences in dose distribution and treatment techniques. Image fusion in this study was completed with rigid registrations; given the time interval between CT simulation and CMR, patient anatomy might have changed. To account for these anatomical changes, rigid registrations included rotations, translations and changes in magnitude aligned to the organ of interest, namely the left ventricle. The power of the current study is limited due to its 28-patient sample size, however, this is comparable to studies listed above. Evaluation of LGE and T1 values on longitudinal studies are expected to further clarify the effect of radiotherapy on myocardial changes.