Abstract
Our purpose was to investigate the utility of 18F-FDG PET/MRI and serial blood work to detect early inflammatory responses and cardiac functionality changes at 1 mo after radiation therapy (RT) in patients with left-sided breast cancer. Methods: Fifteen left-sided breast cancer patients who enrolled in the RICT-BREAST study underwent cardiac PET/MRI at baseline and 1 mo after standard RT. Eleven patients received deep-inspiration breath-hold RT, whereas the others received free-breathing RT. A list-mode 18F-FDG PET scan with glucose suppression was acquired. Myocardial inflammation was quantified by the change in 18F-FDG SUVmean (based on body weight) and analyzed on the basis of the myocardial tissue associated with the left anterior descending, left circumflex, or right coronary artery territories. MRI assessments, including left ventricular functional and extracellular volumes (ECVs), were extracted from T1 (before and during a constant infusion of gadolinium) and cine images, respectively, acquired simultaneously during the PET acquisition. Cardiac injury and inflammation biomarker measurements of high-sensitivity troponin T, high-sensitivity C-reactive protein, and erythrocyte sedimentation rate were measured at the 1-mo follow-up and compared with preirradiation values. Results: At the 1-mo follow-up, a significant increase (10%) in myocardial SUVmean in left anterior descending segments (P = 0.04) and ECVs in slices at the apex (6%) and base (5%) was detected (P ≤ 0.02). Further, a significant reduction in left ventricular stroke volume (−7%) was seen (P < 0.02). No significant changes in any circulating biomarkers were seen at follow-up. Conclusion: Myocardial 18F-FDG uptake and functional MRI, including stroke volume and ECVs, were sensitive to changes at 1 mo after breast cancer RT, with findings suggesting an acute cardiac inflammatory response to RT.
Breast cancer is the most commonly diagnosed cancer and the leading cause of cancer death in women worldwide (1). Adjuvant radiation therapy (RT) of the breast plays a critical role in curative breast cancer management, with local and regional control benefits and lowered mortality rates (2). However, patients with left-sided breast cancer are at an increased risk of radiation-related cardiac disease (3,4), with an increase in the risk of undergoing percutaneous coronary intervention (5) and experiencing cardiac mortality (6), because of the proximity of the heart to the irradiated breast.
A worldwide systematic review on whole-breast RT studies after 2014 reported that the heart received a mean of 3.6 Gy based on 84 left-sided breast cancer studies (7). The left anterior descending coronary artery (LAD), however, had a substantially higher dose than the whole heart, with a mean of 12.4 Gy (7). A population-based case-control study found a linear relationship between major coronary events and the mean heart dose from 2-dimensional breast RT, without a threshold of 7.4% per Gy (8). However, the early effects of radiation are not well understood, and the clinical symptoms do not typically manifest until 10–15 y after RT. Therefore, it is important to limit the exposure of the heart to ionizing radiation during RT to limit the development of cardiac sequelae.
A previous preclinical study of 5 canines imaged with 18F-FDG PET/MRI showed a progressive global inflammatory response during the initial year after RT (9). 18F-FDG PET can identify an inflammatory reaction, because the activated proinflammatory macrophages preferentially sequester glucose. The increased inflammatory uptake was detected as early as 1 wk after single-fraction irradiation of a biologically equivalent LAD dose compared with a standard left breast RT under breath-hold conditions (9). The doses delivered to the whole heart and other coronary arteries were likewise the typical values observed in left-breast RT (7). Immunohistochemistry (CD45) at 12 mo confirmed the presence of inflammatory cells (9).
If inflammation occurs early, preceding but predictive of subsequent cardiac manifestations, there may be a role for early treatment with antiinflammatory or cardioprotective medication. With multimodality imaging, including hybrid PET and MRI, simultaneous acquisition over the same anatomic site allows assessment of acute cardiac inflammation and early functional changes in the heart noninvasively and longitudinally after RT. For optimal 18F-FDG PET assessment of the cardiac inflammatory response, suppression of the normal myocardial uptake of 18F-FDG is required (10).
Functional MRI, including cine imaging, assesses left ventricular (LV) function throughout the cardiac cycle with a short breath-hold of about 15 s. It is considered the gold standard for quantifying LV ejection fraction (LVEF), LV end-diastolic volume, and end-systolic volume (11). In addition, T1 mapping has the ability to detect preclinical myocardial fibrosis. The combination of pre- and postcontrast T1 maps can give a measure of the extracellular volume (ECV), where an increase relates to myocardial fibrosis and correlates with an increasing likelihood of cardiac events (12). The optimal means of quantifying ECV is during a slow, constant infusion of a gadolinium tracer, with a constant concentration of a tracer being supplied to the myocardium during the capture of 3-dimensional T1 maps (13). Lastly, increases in the T2 relaxation rate correlate with an increase in extracellular water, that is, edema.
Serial blood work such as high-sensitivity troponin T (hs-TnT), high-sensitivity C-reactive protein (hs-CRP), and erythrocyte sedimentation rate are the common surrogate markers of myocardial injury and inflammation. The Hs-TnT level has great diagnostic accuracy in detecting acute myocardial infarction (14). Meanwhile, hs-CRP with a level greater than 3 mg/L is associated with higher cardiovascular risk (15). Erythrocyte sedimentation rate can identify acute inflammation by measuring plasma viscosity (16). These biomarkers can provide subclinical evidence of cardiotoxicity during RT.
In this study, we investigated the utility of PET/MRI and serial blood work to detect an early inflammatory response and cardiac functionality changes after RT in patients with left-sided breast RT.
MATERIALS AND METHODS
RT and Delivery
The clinical pilot study (NCT03748030) was approved by the Western University Human Research Ethics Board (HSREB ID 112991), and all subjects gave written informed consent. Of 17 recruited patients, stage T0–T3, 1 patient was ineligible and 1 did not consent. All patients had no prior cardiac disease history, and 1 patient had diabetes mellitus. None of the patients received any prior RT to the thorax or breast.
The patients received their RT during 2020–2021. Most (73%) received standard deep inspiration breath-hold (DIBH) forward-planned intensity-modulated RT (IMRT), 42.5 Gy in 16 fractions, with no adjuvant chemotherapy (67%). Seven of the 11 DIBH RT patients received additional boost doses of 10 Gy in 5 fractions. One patient completed only the first 5 fractions of her RT, discontinuing because of breast swelling, pain, and erythema.
The treatment plans for the 15 patients were retrospectively reviewed. Treatment planning was optimized using the Pinnacle3 treatment planning system (Philips Radiation Oncology Systems). Contours of the heart, left ventricle, and LAD were manually delineated on the treatment planning CT performed on a Brilliance Big Bore CT scanner (Philips) using MIM Maestro (MIM Software Inc.). The mean value for each dose metric is shown in Table 1. This cohort received a low dose in the reported cardiac regions.
Demographics and Radiation Dose Metrics
Imaging
PET/MRI was performed on a 3-T PET/MRI scanner (Biograph mMR; Siemens Medical Systems), with blood work performed at baseline before PET/MRI, within 1 mo after completion of RT, and within 1 y after completion of RT (Fig. 1 shows the protocol). The patients were imaged supine. In this paper, we are reporting the results at the 1-mo follow-up.
PET Imaging (Myocardial Inflammation)
Glycolysis was suppressed through fasting (12 h before imaging) and a 24-h diet high in fat, low in carbohydrate, and low in protein before the PET scan. Furthermore, heparin was injected at 45 min (5 μg/kg) and 30 min (10 μg/kg) before the injection of 18F-FDG. A 60-min list-mode scan of 18F-FDG with a bolus injection at 5 MBq/kg was conducted. All PET data were reconstructed using iterative 3-dimensional ordered-subset expectation maximization (17) with 3 iterations, 21 subsets, 10-min intervals, a 172 × 172 × 127 matrix, and a 4-mm gaussian smoothing filter, yielding a voxel size of 2.08 × 2.08 × 2.03 mm. Attenuation was corrected for all PET scans using a 2-point Dixon MRI pulse sequence (MRI-based attenuation correction), which automatically segments and substitutes discrete attenuation coefficients of the lung, adipose tissue, and soft tissue (18). Myocardial contours were manually delineated on the PET images fused with the MRI-based attenuation correction images using MIM Maestro, according to the American Heart Association 16-segment model (19).
Myocardial inflammation was assessed using the change in the SUVmean (based on body weight) of myocardial tissue from baseline to 40–60 min after tracer injection. SUV at the 1-mo follow-up, compared with baseline, was calculated, with the change being segmented according to coronary artery territory: LAD, left circumflex, or right.
MRI
T2-weighted MR images of the heart using 3 slice locations (apex, mid, and base) were acquired concurrently with PET images using the TrueFISP (Siemens) 2-dimensional sequence with a 224.03-ms repetition time, 1.31-ms echo time, flip angle of 60, field-of-view matrix of 288 × 360, and slice thickness of 6 mm.
The T2-weighted images were followed by a 2-dimensional stack of standard unenhanced steady-state free-precession cine images and T1-weighted images of the whole heart before and during infusion of gadolinium contrast medium (Gadovist; Bayer Inc.). The cine images were acquired using the TrueFISP sequence, a flip angle of 50, a 43.5-ms repetition time, a 1.58-ms echo time, a field-of-view matrix of 253 × 300, and a slice thickness of 6 mm.
The gadolinium contrast medium was injected as a bolus over 2 min (0.1 mmol/kg) followed by a constant infusion over 30 min (0.002 mmol/kg/min). The T1-weighted postgadolinium constant-infusion images were acquired 10 min into the constant infusion. Both sets of T1-weighted images were acquired using the modified look-locker sequence with a 293.92-ms repetition time, 1.22-ms echo time, flip angle of 35, field-of-view matrix of 255 × 300, and slice thickness 6 mm.
Circle CVI42, version 5.11 (Circle Cardiovascular Inc.), was used to assess cardiac function, including LV functional parameters (LV end-diastolic volume, stroke volume (SV), and LVEF), and a radiologist clinically assessed the T2-weighted and T1-weighted postcontrast images. The ECVs were calculated using Equation 1, with extraction of the T1 values of the blood pool and myocardium before and during constant infusion, grouped into 3 slice locations (apex, mid, and basal). The hematocrit ratio was determined from the blood sample.
(Eq. 1)
Statistical Analysis
Statistical analyses were performed using SPSS Statistics for Windows, version 23 (IBM). Shapiro–Wilk normality testing was used to check for normality among the SUVs of 18F-FDG per supplied coronary region, LV functional parameters, blood work, and ECV measurements before and 1 mo after RT. On the basis of the Shapiro–Wilk test, the blood work measurements for hs-TnT, hs-CRP, and erythrocyte sedimentation rate were not normally distributed (P < 0.03). Consequently, tests of significance for these parameters were performed using the Wilcoxon signed-rank test. A paired t test was performed for all other parameters. A Pearson bivariate correlation test was performed to compare these changes to relevant dosimetric parameters of the heart and its substructures (i.e., LV and LAD) presented in Table 1.
Dosimetric parameters of the heart and its substructures were tested for significance between the DIBH and free-breathing-RT group using the Mann–Whitney U test. If any of the changes in 18F-FDG regional uptake, LV functional parameters, blood work, or ECV measurements were significant at follow-up, the Mann–Whitney U test was further performed to check for significance between the DIBH and free-breathing-RT group.
RESULTS
Patient demographics are shown in Table 1. The results for regional uptake of 18F-FDG, LV functional parameters, ECVs, and blood work measurements are presented in Figures 2⇓⇓–5. A significant increase in the average SUVmean in the LAD territory (P = 0.04, 10%) was seen across patients (9/10 patients) at the 1-mo follow-up. A nonsignificant correlation was observed between the increase in uptake in the myocardial tissue region of the LAD territory and the mean and maximum LAD dose metrics, with an r value ranging from −0.23 to −0.24 (P > 0.5). A nonsignificant correlation was observed with the heart dose metrics (mean heart dose and volume of heart receiving at least 5 Gy [V5Gy], with an r value of 0.12–0.17) (P > 0.6) (Tables 2 and 3). The SV was significantly reduced (P < 0.02, 7%, 9/12 patients) at the 1-mo follow-up, whereas LV end-diastolic volume and LVEF did not significantly change (P > 0.08). Most LV functional parameters were within the reference range, except for a single patient who had borderline LV dilation (20). The reduction in SV correlated weakly and insignificantly with all heart and substructure dose metrics (r = 0.14–0.27, P > 0.27). In addition, a significant increase in ECVs in apical and basal slices was identified (P ≤ 0.02 by 6% in 10/12 patients and by 5% in 11/12 patients), whereas no significant change in ECVs was observed for mid slices of the heart (P > 0.5). The ECVs in apical and basal slice locations correlated weak to moderately with all heart, LV, and LAD dose metrics (r = 0.19–0.57, P ≥ 0.07).
Overview of timeline and PET/MRI protocol. GD-DO3A-butrol = gadobutrol; MOLLI = modified look-locker; MRAC = MRI-based attenuation correction.
18F-FDG PET SUVmean of myocardium of 15 patients at baseline and 1-mo follow-up. Uptake values for entire myocardium were categorized by regions supplied by LAD, left circumflex coronary artery, and right coronary artery. Mean value is indicated with +, and median is indicated as median bar. *Significant difference at 1-mo follow-up. LCX = left circumflex coronary artery; RC = right coronary artery; SUVbw = SUVmean based on body weight.
Mean cardiac functional parameters including LV end-diastolic volume, SV, and LVEF for 15 patients before and 1 mo after RT. Significant reduction at 1 mo after RT was shown in SV. Mean value is indicated with +, and median is indicated as median bar. *Significant difference at 1-mo follow-up. EDV = end-diastolic volume; EF = ejection fraction.
Mean ECVs before and 1 mo after RT. Significant increases in ECVs in apical and basal slices were observed at 1-mo follow-up. Mean value is indicated with +, and median is indicated as median bar. *Significant difference at 1-mo follow-up.
Differences in Parameters Between Baseline and 1-Month Follow-up
Presented are mean blood work measurements of hs-TnT, hs-CRP and erythrocyte sedimentation rate before and 1 mo after RT. Mean value is indicated with +, and median is indicated as median bar. ESR = erythrocyte sedimentation rate.
Correlation and Significance of Changes in 18F-FDG PET SUV
No significant changes (P > 0.3) in hs-TnT, hs-CRP, or erythrocyte sedimentation rate measurements were reported. No gross abnormal enhancement, fibrosis, or edema measured with T1- and T2-weighted images at either baseline or the 1-mo follow-up was detected on review by a cardiac radiologist. One patient had borderline LV dilation at the 1-mo follow-up.
For dose metrics, only the mean heart dose (P = 0.04) was significantly higher in free-breathing RT than in DIBH RT patients. The maximum heart and LAD dose, heart V5Gy, mean LV, and LAD dose did not significantly differ between the RT groups (P ≥ 0.06). Changes in uptake in the LAD territory, SV, and ECVs in apical and basal slices did not significantly differ (P ≥ 0.2) between the RT groups.
DISCUSSION
Currently, dose-sparing guidelines for cardiac substructures are not well established in breast RT. In terms of the whole heart, consensus guidelines recommend that the volume of the heart irradiated should be minimized as much as possible without compromising the breast target coverage. The guideline called “Quantitative Analyses of Normal Tissue Effects in the Clinic” recommended limiting the volume of heart receiving at least 25 Gy to less than 10% to keep the risk of cardiac mortality under 1% (21). In our study, the cardiac doses among patients were lower than the guideline, with a mean whole-heart V5Gy of 9.46%. However, the LAD and left ventricle can still receive a substantially higher dose than the remainder of the heart structures (7). The mean LAD dose in our study was 2.78 Gy, which was recognized as a region with a high dose compared with the overall heart (mean dose, 1.79 Gy; Table 1).
Although cardiac risk reduction strategies such as the role of active-breathing modalities (22), patient positioning (23,24), and accelerated partial breast irradiation (25) are discussed, few efforts in randomized controlled trials have validated the cardiac-sparing techniques or looked into the early response of cardiac substructures to breast RT. The fact that 3 patients received a higher mean heart dose in free-breathing RT but showed no differences in other dosimetric parameters of the heart and its substructures implies that DIBH RT can achieve better sparing in terms of the whole heart than free-breathing RT can. With PET/MRI, the significantly elevated uptake of 18F-FDG in LAD segments, along with the increase in ECVs in apical and basal slices with a reduction in SV, suggested acute regional inflammation or functional changes in the myocardium as early as 1 mo after the end of RT. The changes were observed in patients even with low-dose myocardial irradiation, compared with the recommended guidelines, regardless of breath-hold techniques.
Jo et al. (26) conducted a retrospective study evaluating irradiated myocardium segmented on the basis of dose threshold in both staging and post-RT PET/CT images of breast cancer patients who underwent 3-dimensional CRT. Uptake in myocardium irradiated with more than 30 Gy significantly increased after RT even at the 1-y follow-up. The degree of uptake increase significantly correlated with the radiation dose to the myocardium. However, glucose suppression was not performed. In our study, in which glucose was suppressed and the radiation dose to the myocardium was low, the uptake increase in the LAD segments correlated weakly with the whole-heart dose metrics. Also of note, the myocardium was segmented according to the AHA heart model, which can better locate the radiosensitive substructure of the heart, that is, the LAD myocardial segments.
In terms of MRI functional parameters, our study demonstrated a significant reduction in SV at the 1-mo follow-up; no significant changes were shown in LV end-diastolic volume and LVEF. This corresponds to the results of a systematic review by Kaidar-Person et al. (27), which reported 5 of 6 studies without LVEF reduction using SPECT imaging at the 6-mo follow-up and 4 studies with perfusion defects. Bergom et al. (28) evaluated ECVs and LV functional parameters in a pilot study of breast cancer patients who received 3-dimensional conformal RT and adjuvant anthracycline-based chemotherapy using cardiac MRI. No clinically abnormal findings were seen at a median follow-up of 8.3 y after RT, and no evidence of increased ECVs with increasing heart or ventricular radiation doses was reported (28). That finding is contrary to our study, which identified a weak to moderate correlation between the increase in ECVs (at apical and basal slice locations) and the heart and substructure metrics. However, this study performed only a median long-term follow-up scan; hence, changes in the LV functional parameters and ECVs were not determined. Without measurements performed before 6 mo, any early postulated effects of radiation on myocardial metabolism are purely conjecture.
Limitations of our pilot study include a small sample size of 15 patients, of whom 2 had insufficient glucose suppression in their baseline 18F-FDG PET scan and 2 did not complete the 1-mo post-RT imaging. However, in the literature, suppression fails 5% of the time even under the best diet and fasting protocols (29). The sample size of patients between breath-hold and free-breathing RT techniques was small; hence, in future a larger sample is needed to increase the power of comparison of early cardiac response between RT techniques. Furthermore, hs-TnT, hs-CRP, and erythrocyte sedimentation rate are early cardiac inflammation biomarkers that may return to normal values by 1 mo unless persistent myocardial injury is present; hence, these biomarkers should be assessed at earlier and later time points.
It is unlikely that the 70-min PET/MRI protocol used in our study would be routinely used for patient management. Furthermore, within 1 mo after RT, none of the patients had clinically significant cardiac events; therefore, we do not recommend that these findings influence present clinical practice. However, scarring can manifest at a later stage, such that additional care to minimize the volume of cardiac substructure (LAD or left ventricle) in the RT field and longitudinal follow-up are recommended. With patients returning for their 1-y post-RT imaging, longitudinal 1-y follow-up would increase the power to detect transformation of subsequent inflammation changes into cardiac sequelae such as progressive fibrosis or scar formation. Such evidence-based information can help establish guidelines to determine the need for cardiovascular risk assessment in patients before initiation of RT and long-term cardiovascular monitoring of breast cancer survivors, in addition to the modification of the cardiovascular risk-based RT regimen.
CONCLUSION
We observed a small but statistically significant increase in 18F-FDG uptake in the myocardial territory of the LAD, along with a significant increase in ECVs at the apical and basal locations of the heart, at 1 mo after the end of left-sided breast cancer RT in a small cohort. These observations may be related to a significant decrease in the LV SV noted at follow-up. No significant changes in blood work measurements, including hs-TnT, hs-CRP, and erythrocyte sedimentation rate, were seen. Recognizing that the absolute change in SUV may be small in a small cohort of patients, the clinical significance of this change has yet to be realized. Among the patients, our pilot study demonstrated the feasibility of using PET/MRI to assess the cardiac response to RT as early as 1 mo of follow-up. Validation of these metrics in the prediction of radiation-induced cardiac disease in a larger cohort might prompt a change in management of left-sided breast cancer patients with early cardiac changes detected with noninvasive imaging.
DISCLOSURE
This clinical pilot study (NCT03748030, November 16, 2018) was approved by the Western University Human Research Ethics Board (HSREB ID 112991), funded by the Lawson Strategic Research Fund (R19-292). This work was partly funded by the Translational Breast Cancer Studentship Award from the Breast Cancer Society of Canada. No other potential conflict of interest relevant to this article was reported.
KEY POINTS
QUESTION: Are both 18F-FDG PET/MRI and serial blood work capable of detecting early inflammatory responses and cardiac functionality changes at 1 mo after RT in patients with left-sided breast cancer?
PERTINENT FINDINGS: 18F-FDG/PET myocardial uptake and functional MRI, including SV and extracellular matrix measurements, were sensitive to changes at 1 mo after breast cancer RT, with findings suggesting an acute inflammatory response in the heart.
IMPLICATIONS FOR PATIENT CARE: If inflammation occurs early, preceding but predictive of subsequent cardiac manifestations, there may be a role for early treatment with antiinflammatory or cardioprotective medication.
ACKNOWLEDGMENTS
We acknowledge the participating clinical trial patients and their families.
Footnotes
Published online May 16, 2023.
REFERENCES
- Received for publication September 26, 2022.
- Revision received March 30, 2023.
