Presented in part at the 2004 Annual Meeting of the American Society for Clinical Oncology, New Orleans, Louisiana, June 5–8, 2004 and at the 46th Annual Meeting of the American Society for Therapeutic Radiology and Oncology, Atlanta, Georgia, October 3–7, 2004.
Radiation therapy (RT) to the left breast/chest wall has been linked with cardiac dysfunction. Previously, the authors identified cardiac perfusion defects in approximately 50% to 60% of patients 0.5 to 2 years post-RT. In the current study, they assessed the persistence of these defects 3 to 6 years post-RT.
From 1998 to 2006, 160 patients with left-sided breast cancer were enrolled onto an Institutional Review Board-approved, prospective study. All patients received tangential photons to the left breast/chest wall. Patients had pre-RT and serial post-RT single-photon emission computed tomography (SPECT) scans to assess changes in regional cardiac perfusion, wall motion, and ejection fraction (EF). Forty-four patients had SPECT scans 3 to 6 years post-RT and were evaluable for the current analysis.
The overall incidence of perfusion defects at 3 years, 4 years, 5 years, and 6 years was 52% (11 of 21 patients), 71% (17 of 24 patients), 67% (12 of 18 patients), and 57% (4 of 7 patients), respectively. The rate of abnormal SPECT scans 3 to 6 years post-RT in patients who had scans at 0.5 to 2 years that were either all abnormal, intermittently abnormal, or all normal was 80%, 67%, and 63%, respectively. The incidence of wall motion abnormalities in patients with or without perfusion defects 3 to 6 years post-RT was low and did not differ statistically (17% vs 7.1%, respectively; P = .65), as was the incidence of reductions in EF of ≥5% (27% vs 36%, respectively; P = .72).
Radiation therapy (RT) plays an important role in the treatment of patients with breast cancer, in both the postlumpectomy and postmastectomy settings. In both of these situations, the addition of RT clearly improves locoregional control and also appears to improve overall survival.1, 2 However, the survival gains associated with improvements in locoregional control may be diminished by RT-associated cardiac mortality. Several studies have demonstrated that the reductions in cancer-specific deaths afforded by RT are off-set in part by increased cardiac mortality.3, 4
The risk of cardiac dysfunction post-RT for breast cancer appears to be related to the RT techniques used. Older RT techniques, which resulted in incidental irradiation of large volumes of the heart to relatively high doses, have been associated with an increased risk of cardiac morbidity and mortality.3–6 The results from studies that examined patients who were treated with more modern RT techniques (which tend to limit cardiac exposure) have been conflicting. Several studies have suggested an increased risk of coronary artery disease and myocardial infarction in patients treated with relatively modern RT techniques.7, 8 In contrast, several other studies of patients treated with modern RT techniques have not demonstrated an increase in cardiac events.9–13 One important caveat is that the follow-up in studies of patients treated with modern techniques generally is considerably shorter than in studies of patients treated with older techniques.
We previously reported that RT delivered using modern techniques resulted in myocardial perfusion defects in approximately 50% to 60% of patients. The incidence of these defects appears to be related to the volume of left ventricle (LV) irradiated, with an incidence of < 20% in patients with < 1% of the LV irradiated versus an incidence from > 70% to 80% in patients with ≥5% of the LV radiated.14 Similar but smaller studies conducted by other groups have generated results that largely are consistent with our own.15, 16
The clinical and functional significance of these perfusion defects is unclear. In our experience, the presence of perfusion defects has been associated with abnormalities of wall motion, subtle reductions in ejection fraction (EF), and episodes of chest pain (of unclear etiology, but probably related to pericarditis).14, 17–19 In similar studies, others have also noted wall motion abnormalities that were associated with perfusion defects.17
Although some retrospective functional cardiac imaging studies have examined patients at long intervals post-RT, the prospective studies published to date, including our own, have evaluated patients at relatively short post-RT intervals. In the current article, we report the results of myocardial perfusion imaging obtained in our prospectively studied cohort of breast cancer patients from 3 to 6 years after treatment.
MATERIALS AND METHODS
From 1998 to 2006, 160 patients with left-sided breast cancer were enrolled onto an Institutional Review Board-approved, prospective study to assess RT-induced cardiac dysfunction. Patients had pre-RT and serial post-RT single-photon emission computed tomography (SPECT) scans to assess changes in regional cardiac perfusion, wall motion, and EF. The study was designed initially to image patients every 6 months for up to 2 years post-RT. Limited funds were secured later to study some of the patients ≥3 years post-RT. Because funds were limited, initially, we preferentially recruited patients who had abnormal scans at 0.5 to 2 years post-RT. Later, we recruited patients who had normal scans at 0.5 to 2 years post-RT for evaluations at ≥3 years. Figure 1 shows how the current study cohort of 44 patients was derived from the total population of 160 patients enrolled on the study.
For the current analysis, we examined the subset of 44 patients who had follow-up evaluations 3 to 6 years post-RT and had normal baseline (pre-RT) SPECT scans. Twenty of those 44 patients had only a single scan during this period. Twenty-two patients were scanned twice during the 3- to 6-year follow-up period, and 2 patients were scanned 3 times.
Detailed descriptions of the RT techniques and the methodology for 3-dimensional (3D) radiation dose calculation, performance and interpretation of cardiac SPECT scans, registration of SPECT scans with the radiation dose distribution, and statistical analyses are provided elsewhere.14 In brief, all patients received tangential photons to the left breast or chest wall to a planned dose of 46 to 50 gray (Gy) at 1.8 to 2.0 Gy per fraction; 1 patient only received 44 Gy due to acute skin toxicity. The regional lymph nodes, generally including the internal mammary lymph nodes (IMNs), were irradiated at the discretion of the treating physician. Twenty-one of 44 patients had their IMNs treated, including 19 patients who were treated using partially wide tangents and 2 patients who were treated using a separate, mixed-beam IMN field. Patients who had RT after mastectomy received an additional 10-Gy electron boost to the mastectomy scar. Patients who were irradiated after breast-conserving surgery received a 16- to 20-Gy electron boost to the tumor bed. Exit doses to the heart from the electron component of the IMN treatment and the tumor bed/mastectomy scar boost (both administered with en face electrons) were not considered in our calculations, because electron dosimetry was not available at the time of this analysis.
3D Radiation Dose Calculation
A pre-RT planning computed tomography (CT) scan was performed on each patient in the treatment position using a dedicated CT scanner in the Department of Radiation Oncology. The inhomogeneity-corrected 3D radiation dose distribution was computed using Plan University North Carolina (PLUNC) treatment planning software.20
Quantitative Assessment of Cardiac Function
Cardiac SPECT scans were used to provide objective quantitative data regarding LV regional myocardial perfusion, regional wall motion, and EF. Although the other cardiac chambers may be exposed to radiation in some circumstances (eg, patients treated with a separate IMN field), cardiac SPECT scans image only the LV. A nuclear medicine radiologist who was blinded to clinical information scored each patient's SPECT scans independently. A 12-segment reporting system was used to quantify both perfusion and wall motion. EF was computed automatically from the SPECT scan data using accepted techniques.21–23
Registration of SPECT Scans and Radiation Dose Distribution
The pre-RT and serial post-RT SPECT perfusion scans were performed with patients in the treatment position (supine, arms up). The pre-RT SPECT perfusion images then were registered to the treatment-planning CT images.24 The LV was contoured on the merged CT/SPECT images, and the percentage of the CT/SPECT-defined LV that received different doses of radiation was then computed. Figure 2 shows the tangential RT fields superimposed on the registered CT/cardiac SPECT image from a representative patient and illustrates the typical volume of LV within the RT fields.
Endpoints and Statistics
The following endpoints were considered in the current study: 1) the incidence and predictors of RT-associated perfusion defects at discrete time intervals from 3 to 6 years post-RT, 2) the incidence of wall motion abnormalities 3 to 6 years post-RT, and 3) the incidence of ≥5% reductions in EF 3 to 6 years post-RT. Incidence rates were compared using a 2-tailed Fisher exact test. A P value of .05 was considered significant. Predictors of perfusion defects were assessed using logistic regression. To determine whether the severity of the perfusion defect had an impact on the incidence of wall motion abnormalities or the incidence of decreases in EF ≥5%, we examined the incidence of these 2 endpoints as a function of the Summed Rest Score (SRS) using logistic regression. The SRS is determined by rating the extent of perfusion defects in each of 12 segments of the heart on a scale from 0 to 4. Accordingly, scores on the SRS can range from 0 to 36.
The baseline patient, tumor, and treatment characteristics of the study population are shown in Table 1. Of the 44 evaluable patients, 21 patients, 24 patients, 18 patients, and 7 patients had SPECT scans obtained at 3 years, 4 years, 5 years, and 6 years post-RT, respectively. The median follow-up was 4.3 years (range, 2.5–7 years). Twenty patients had only a single scan from 3 to 6 years after treatment; 22 patients had scans at 2 time points, and 2 patients had scans at 3 time points.
BMI indicates body mass index; RT, radiotherapy; LV, left ventricle; Gy, grays; IMNs, internal mammary nodes; SPECT, single photon emission computed tomography.
All values indicate the number of patients unless stated otherwise.
Median age, y, (range)
Cardiac risk factor
Coronary artery disease
Obesity, BMI ≥30 kg/m2
Mean RT dose to breast/chest wall, Gy
Mean percentage of LV in RT field
Separate IMN field
Partly wide tangents
SPECT results 0.5–2 y after RT
All normal scans
Incidence of Perfusion Defects 3 to 6 Years Post-RT
Among the 44 evaluable patients, the overall incidence of perfusion defects at 3 years, 4 years, 5 years, and 6 years was 52%, 71%, 67%, and 57%, respectively. For the entire 3- to 6-year period, the crude rate of perfusion defects was 68%. In these selected patients (most of whom developed perfusion defects 0.5–2 years post-RT), the cumulative incidence of ever having a perfusion defect was 60% at 3 years and 90% at 6 years (the intervals from 0.5 to 3 years and from 0.5 to 6 years, respectively, were considered).
The incidence of perfusion defects (defined as an SRS > 0) during 3 to 6 years, stratified according to each patient's 0.5- to 2-year scan results, is shown in Table 2. Twelve of 19 patients (63%) who had all normal scans during the first 2 years of follow-up had new perfusion defects noted during years 3 to 6, and the median time to identification of a new defect of 47 months. Eight of 10 patients (80%) who had all abnormal scans 0.5 to 2 years post-RT continued to have abnormal scans 3 to 6 years post-RT. Finally, 10 of 15 patients (67%) who had intermittently abnormal scans during early follow-up (ie, some normal scans and some abnormal scans during the ≤2-year interval) continued to have abnormal scans with extended follow-up. In multivariable analysis, after controlling for the percentage of LV in the RT field, the 0.5- to 2-year scan result had no predictive value for the outcome of developing a perfusion defect at 3 to 6 years. The percent LV irradiated was the only predictive factor identified in our current and prior multivariate analysis.24 However, the number of evaluable patients in each subgroup is small; thus, the power to detect differences in subgroups is limited.
Table 2. Rates of Perfusion Defects 3 to 6 Years After Radiotherapy Stratified by 0.5- to 2-Year Single-photon Emission Computed Tomography Scan Results
Scan results 0.5–2 y Post-RT
Incidence of perfusion defects on later scans (3–6 y Post-RT) no. of patients/total no. (%)
The total at 3 to 6 years does not equal the sum at each individual time point, because some patients had more than 1 scan obtained from 3 to 6 years after RT. If a patient had both normal and abnormal scans from 3 to 6 years after RT, then the patient was scored as having a perfusion defect for the cumulative results (far right column of the table).
All normal, n = 19
All abnormal, n = 10
Intermittent defects, n = 15
In general, the perfusion defects that were observed 3 to 6 years post-RT occurred in the same location as the defects that were detected 0.5 to 2 years post-RT, namely, in the anterior segments of the LV (segments that typically lie within the tangential breast/chest wall RT fields), as described in Tables 3 and 4. Only 3 of 44 patients developed any clinical cardiac complications; all 3 of those patients had acute pericarditis post-RT.
Table 3. Locations of Late (3–6 Years Postradiotherapy) Versus Early (0.5–2 Years Postradiotherapy) Perfusion Defects Stratified by 0.5- to 2-Year Scan Results
0.5- to 2-Year scan results
No. of patients/total no.
Location of perfusion defects 3–6 years post-RT: no. of patients/total no. (%)*
Defects in same location(s) noted 0.5–2 y post-RT
Defects in new location(s)
Defects in old and new location(s) noted 0.5–2 y post-RT
RT indicates radiotherapy.
Not all defects that developed 3 to 6 years after RT persisted.
Table 4. Location of Perfusion Defects 3 to 6 Years After Radiotherapy
Incidence of Wall Motion Abnormalities 3 to 6 Years Post-RT
The incidence of wall motion abnormalities 3 to 6 years post-RT in patients with and without perfusion defects in the same post-RT interval is shown in Table 5. All wall motion abnormalities were classified as hypokinetic and typically involved a small portion of the anterior and apical regions of the LV. The incidence of wall motion abnormalities in patients with and without perfusion defects 3 to 6 years post-RT was low in both groups and did not differ statistically. The location of the wall motion abnormalities corresponded to the location of the perfusion defects in 4 of 7 scans.
Table 5. The Incidence of Wall Motion Abnormalities Detected on Cardiac Single-photon Emission Computed Tomography Scans Performed 3 to 6 Years After Radiotherapy in Patients With and Without Perfusion Defects
Incidence of wall motion abnormalities: no. of patients/total no. (%)
The incidence of EF declines ≥5% that occurred 3 to 6 years post-RT in patients with and without perfusion defects during the same post-RT interval is shown in Figure 3. The incidence of reductions in EF of ≥5% in patients with and without perfusion defects 3 to 6 years post-RT was modest and did not differ statistically (27% vs 36%, respectively; P = .72). The correlation between change in EF and the severity of the perfusion defect, as quantified using the SRS, is shown Figure 4. Higher SRS values were not correlated with reductions in EF ≥5% (P = .75).
Analysis of the Impact of Selection Bias
It is likely that there were differences in the characteristics of patients who had cardiac scans obtained 3 to 6 years post-RT compared with the larger group of patients who had scans obtained 0 to 2 years post-RT. Because of funding restrictions, we did not try to rescan all patients systematically during the latter time points. Rather, we preferentially encouraged patients who had abnormalities on 1 or several of their scans in the 0.5- to 2-year interval to return for scans at a later time. When funds were acquired to pay for scans at the later follow-up points, letters were sent to these patients to encourage them to return for additional evaluation.
Several factors probably had an impact on the decision of whether a patient returned for follow-up scans. These included the patient's prior scan results (previous abnormalities), distance to our facility, and the patient's disease status (recurrence-free or not). In addition, other unknown factors may have had an impact a patient's decision.
To explore the issue of selection bias further, we compared the distribution of prognostic factors (established in our prior publications14, 24) for the development of perfusion defects in patients who were included in the current study versus those who were eligible for inclusion but were not evaluable for the current study (Table 6). Given our preferential enrollment of patients with perfusion defects, as expected, the percentage of LV in the treatment field was significantly higher in patients who were evaluable for the current analysis versus those who were eligible but did not have scans obtained 3 to 6 years post-RT. However, the mean body mass index was actually significantly lower in the patients who were evaluable in the current study compared with those who were eligible but did not undergo scans at the 5- and 6-year follow-up intervals.
Table 6. Summary of Patient-specific Factors Among Patients Who Returned for Follow-up 3 to 6 Years After Treatment and Patients Who Were Eligible but Did Not Return for Follow-up
In our selected group of patients who received RT for left-sided breast cancer, we observed a high incidence of perfusion defects 3 to 6 years after treatment. The incidence of perfusion defects 3 to 6 years post-RT in the broader population of patients undergoing RT for left-sided breast cancer is unknown but probably is lower. In addition, the functional significance of these perfusion abnormalities remains unclear, because the perfusion defects we observed were not associated with wall motion abnormalities or with significant changes in EF.
Six studies, 5 of which were retrospective, have previously examined the potentially cardiotoxic effects of RT for left-sided breast cancer using myocardial perfusion imaging (Table 7). The studies in which patients were irradiated using techniques similar to our own and studied in a similar time frame post-RT reported high rates of perfusion defects, consistent with the results reported herein. The studies in which patients were irradiated using different techniques or studied in a markedly different time frame reported lower rates of perfusion defects.
Table 7. Summary of Myocardial Perfusion Scintigraphy Studies Assessing Cardiac Injury in Patients With Breast Cancer
Two studies, 1 by Seddon et al. and 1 by Gyenes et al., used techniques similar to ours and studied patients within a similar time frame. In a retrospective study, Seddon et al. reported LV perfusion defects in 17 of 24 patients who had left-sided breast cancer (71%) compared with 2 of 12 patients who had right-sided breast cancer (17%) who were irradiated ≥5 years before imaging (P = .002).16 In addition, there was a higher rate of regional myocardial wall motion abnormalities in patients with left-sided versus right-sided tumors (35% vs 0%, respectively; P = .071). Similar to our study, no significant reductions in EF were reported in either group of patients.
In the only prospective study (other than our own) conducted to date, Gyenes et al. reported perfusion defects in 6 of 12 patients with left-sided breast cancer (50%) approximately 1 year post-RT.15 The RT technique appeared to have a significant impact on the subsequent risk of cardiac injury; 4 of 4 patients who were treated with tangential photons developed a perfusion defect versus 2 of 8 patients who were treated with en face electrons.
Four functional imaging studies have been conducted in which either the RT technique or the length of follow-up differed substantially from our own. Three of those studies (Hojris et al., Gustavvson et al., Cowen et al.) did not detect an increased rate of perfusion defects; a fourth study (by Gyenes et al.) identified perfusion defects but at a substantially lower rate than in our current study. Hojris et al. performed rest and stress myocardial perfusion imaging studies in 17 patients who had taken part in 1 of 2 Danish randomized trials of postmastectomy RT (Danish Breast Cancer Cooperative Group studies 82b and 82c) between 1982 and 1990.25 The scans were obtained a median of 8 years after treatment. Perfusion defects were observed in 4 of 9 irradiated patients (44%) and in 4 of 7 nonirradiated patients (57%). None of the defects in the irradiated patients were located in the anterior segment of the heart.
There are several reasons why Hojris et al. may not have observed an increased rate of perfusion defects. First, the defect rate in the unirradiated patients was surprisingly high (57%), which may have made it difficult to detect a difference between the control group and the treated patients. Second, the RT techniques used by Hojris et al. (en face electrons) may have been less cardiotoxic than the techniques used in our patients (partially wide tangents or a separate IMN field matched to shallow tangents). Third, in contrast to the Danish trials, in which combined cyclophosphamide, methotrexate, and 5-fluorouracil chemotherapy was used, many of our patients received doxorubicin-based chemotherapy, which may potentiate the cardiotoxic effects of radiation.26
Gustavsson et al. performed myocardial scintigraphy and echocardiography/Doppler ultrasound 10 to 17 years post-RT in 90 women who were treated on the South Sweden Breast Cancer Trial, a trial in which women with stage II breast cancer were randomized after mastectomy to receive RT, RT with chemotherapy (cyclophosphamide), or chemotherapy alone.27 The rate of perfusion defects in that trial was 4 of 34 in patients who were irradiated for left-sided tumors, 2 of 33 in patients who were irradiated for right-sided tumors, and 0 of 23 in patients who received chemotherapy only; these differences were not statistically significant.
Several factors may explain the discrepancy between our results and those reported by Gustavsson et al. First, as in the study by Hojris et al., the RT techniques used by Gustavsson et al. differed significantly from our techniques: The IMNs were treated exclusively with electrons, and the medial chest wall was treated with orthovoltage x-rays, techniques that reduce the exit dose to the heart but have other drawbacks. Second, the patients either received no chemotherapy or received chemotherapy consisting only of cyclophosphamide. In contrast, 68% of our patients received doxorubicin-based chemotherapy.
Cowen et al. performed exercise thallium-201 scans in 17 patients with left-sided breast cancer an average of 4.6 years post-RT. The patients were treated from 1987 to 1993, and all received IMN radiation using a 2:1 ratio of photons to electrons.28 No significant abnormalities were demonstrated in any of the patients on either exercise-tolerance testing or myocardial perfusion imaging.
There are several potential explanations why Cowen et al. may not have observed any cardiac abnormalities. First, the RT techniques used by Cowen et al. (incorporating a separate IMN field) may be less cardiotoxic than those used in our study (generally, partially wide tangents). The comparative cardiac toxicity has not been studied well for these 2 approaches. Second, the nuclear medicine imaging techniques used were slightly different from our own.
In 1 retrospective study, perfusion defects were observed post-RT for left-sided breast cancer but at a lower rate than we observed in the current study. Gyenes et al. performed exercise cardiac SPECT scans and echocardiography in 20 patients with left-sided breast cancer and in 17 right-sided controls (no RT or right-sided RT) an average of 18.4 years (left-sided patients) and 19 years (right-sided patients) post-RT.29 (These patients had been treated from 1971 to 1976 on the Stockholm Breast Cancer Trial.) Perfusion defects were observed in 5 of 20 patients with left-sided tumors (25%) and in 0 of 17 patients with right-sided tumors (0%). All 5 perfusion defects were located in the anterior myocardium. None of the perfusion defects were associated with abnormal LV EF or wall motion.
There are several reasons why Gyenes et al. may have observed a lower rate of perfusion defects than we observed. First, none of the patients in the study by Gyenes et al. received adjuvant chemotherapy. Second, no attempt was made to determine whether any heart actually was in the RT fields by reviewing simulation or port films. Third, in 9 of 20 of the patients with left-sided breast cancer, the chest wall was treated with en face electrons instead of tangential photons, as was done in the current study. Finally, the interval between RT and imaging in their study was much longer than in ours (average, 18.4 years in the study by Gyenes et al. vs a maximum of 6 years in the current study). It is possible that some of the perfusion defects we observed may dissipate over time.
In summary, studies that have examined RT-associated cardiac toxicity using myocardial imaging have demonstrated perfusion defects in 0% to 70% of patients from 1 to 18 years post-RT for left-sided breast cancer. This wide variability in the incidence of perfusion defects post-RT probably was caused by the variable RT techniques used to treat patients in different studies, the variable use of adjuvant systemic (potentially cardiotoxic) chemotherapy, differing patient selection criteria and biases, and differing intervals between treatment and subsequent perfusion imaging. The results from the current study and from the other published series of myocardial perfusion imaging in patients who were irradiated for early-stage breast cancer are summarized in Table 7.
Our study had several limitations. First, although, to our knowledge, we have accumulated the largest cohort of patients prospectively evaluated for cardiac toxicity post-RT for left-sided breast cancer, the number of patients with follow-up > 2 years remained relatively small. Second, selection bias may have influenced our results. Because of funding limitations, initially, we preferentially recruited patients who had perfusion defects identified 0.5 to 2 years post-RT to undergo repeat cardiac SPECT scans 3 to 6 years post-RT. Later, we attempted to obtain 3- to 6-year scans in patients who had no perfusion defects identified in the first 2 years post-RT. However, physician and patient selection bias likely had a significant impact on recruitment within this subgroup of patients. Third, although it has been demonstrated that cardiac SPECT predicts cardiac morbidity and mortality in patients with ischemic heart disease,30 the clinical significance of abnormal cardiac SPECT scans in irradiated breast cancer patients is unknown. Fourth, as we reported previously, the visual scoring of the SPECT scans performed in this study was subjective and may have underestimated the true incidence of abnormalities.31 Fifth, because SPECT scans image only the LV, they do not detect RT-induced injury in the other cardiac chambers. Finally, as we also reported previously, clinical (patient-specific) factors, such as diabetes, hypertension, and body mass index, may mitigate or exacerbate the effects of RT on the heart.24 Unfortunately, in the current study, small sample size precluded us from examining the impact of these factors.
In conclusion, the results from prior studies indicate that RT causes perfusion defects, wall motion abnormalities, and subtle changes in EF from 0.5 to 2 years post-RT. Our current study suggests that these perfusion defects are likely to persist from 3 to 6 years post-RT. Furthermore, new perfusion defects may appear 3 to 6 years post-RT in patients who had normal cardiac SPECT scans 0.5 to 2 years after treatment. Although we were unable to detect an increase in the risk of functional cardiac abnormalities 3 to 6 years post-RT, we believe that additional follow-up is needed to gain a better understanding of the persistence and functional consequences of cardiac dysfunction in patients who receive RT for breast cancer. In the meantime, every effort should be made to minimize incidental irradiation of the heart while maintaining adequate coverage of target volumes.
We thank Phil Antoine for help with data storage and analysis, Robert Pagnanelli for assistance in processing the cardiac images, and the University of North Carolina at Chapel Hill for the use of PLUNC treatment-planning software.