mortality from myocardial infarction following ... - CiteSeerX

Int. J. Radiation Oncology Biol. Phys., Vol. 43, No. 4, pp. 755–761, 1999 Copyright © 1999 Elsevier Science Inc. Printed in the USA. All rights reserved 0360-3016/99/$–see front matter

PII S0360-3016(98)00412-X

CLINICAL INVESTIGATION

Breast

MORTALITY FROM MYOCARDIAL INFARCTION FOLLOWING POSTLUMPECTOMY RADIOTHERAPY FOR BREAST CANCER: A POPULATION-BASED STUDY IN ONTARIO, CANADA LAWRENCE F. PASZAT, B.A., M.D., M.S., F.R.C.P.C.,*† WILLIAM J. MACKILLOP, M.B.CH.B., F.R.C.P.C., F.R.C.R.,*† PATTI A. GROOME, PH.D.,*† KARLEEN SCHULZE, B.MATH., M.MATH.,* AND ERIC HOLOWATY, M.D., M.SC., F.R.C.P.C.‡ *Radiation Oncology Research Unit, Department of Oncology, Queen’s University and Kingston Regional Cancer Centre, Kingston, Ontario, Canada; †Department of Community Health and Epidemiology, Queen’s University, Kingston, Ontario, Canada; and ‡Ontario Cancer Registry and the University of Toronto, Toronto, Ontario, Canada Purpose: To compare the risk of mortality from myocardial infarction (MI) after left-sided postlumpectomy radiotherapy (RT) to the risk after right-sided postlumpectomy RT. Methods: We conducted a population-based cohort study of cases of invasive female breast cancer in Ontario, diagnosed between January 1, 1982 and December 31, 1987 (n 5 25,570). Records of the Ontario Cancer Registry (OCR) were linked to hospital procedure and discharge abstracts and to RT records from Ontario cancer centers. A case was labelled as lumpectomy if this was the maximum breast surgery within 4 months of diagnosis. Postlumpectomy RT occurred up to 1 year postdiagnosis. Laterality was assigned from the laterality descriptor of the RT records. A case was labelled as having had a fatal MI if ICD code 410 (myocardial infarction) was recorded as the cause of death in the OCR. We used logistic regression to compare the likelihood of utilization of : 1. Dose per fraction > 2.00 Gy; 2. cobalt vs. linac; and 3. boost RT. We used life table analysis and the log rank test comparing the time to fatal MI from diagnosis of breast cancer between women who received left-sided postlumpectomy RT and women who received right-sided. We used Cox proportional hazards models to study the relative risk for left-sided cases overall, and stratified by age, RT characteristics, and among conditional survival cohorts. Results: Postlumpectomy RT was received by 1,555 left-sided and 1,451 right-sided cases. With follow-up to December 31, 1995, 2% of women with left-sided RT had a fatal MI compared to 1% of women with right-sided RT. Comparison of the time to failure between women who had left-sided RT and women who had right-sided RT showed the left-sided RT group to be associated with a higher risk of fatal MI (p 5 0.02). Adjusting for age at diagnosis, the relative risk for fatal MI with left-sided postlumpectomy RT was 2.10 (1.11, 3.95). Conclusion: Among women who received postlumpectomy RT for breast cancer in Ontario between 1982–1987, left-sided postlumpectomy RT was associated with a higher risk of fatal MI compared to right-sided. © 1999 Elsevier Science Inc. breast cancer, Radiotherapy, Myocardial infarction.

INTRODUCTION Associations between adjuvant radiotherapy (RT) for breast cancer and increased mortality have been described recently; however, it has been unknown if this association exists among women who have received postlumpectomy RT. A report from the Early Breast Cancer Trialists’ Cooperative Group demonstrated an increased risk of mortality from causes other than breast cancer among women who had received adjuvant RT (1). Previous meta-analysis of clinical trials of postmastectomy RT for breast cancer between 1949 and 1975 demonstrated an increased risk of mortality from cardiac causes among irradiated cases (2). A

population-based study of the Swedish cancer registry found an increased risk of mortality from myocardial infarction (MI) among cases of cancer of the left breast compared to cancer of the right breast between 1970 and 1985; however, details of surgery, and the proportion of women who received adjuvant RT were unknown (3). We have reported an overall relative risk of fatal MI of 1.17 after adjuvant RT for left-sided breast cancer compared to right-sided breast cancer in the Surveillance, Epidemiology and End Results (SEER) registry population from 1973– 1992 (4). It is unclear if the associations between RT and cardiac

Reprint requests to: Lawrence F. Paszat, M.D., Radiation Oncology Research Unit, Kingston General Hospital, Apps Level 4, Kingston, Ontario K7L 2V7 Canada. Acknowledgements—Supported by a grant from Cancer Care On-

tario. P. A. G. is a career scientist of the Ministry of Health of Ontario, Canada. We thank the radiation oncology departments of Ontario who provided access to their records. Accepted for publication 20 September 1998. 755

756

I. J. Radiation Oncology

●

Biology

●

Physics

mortality found in studies involving large numbers of cases treated with postmastectomy RT prior to the lumpectomy era may be generalized to the population of women undergoing lumpectomy. On the one hand, RT practices in the lumpectomy era have changed from those represented in the meta-analyses. Internal mammary nodal RT via a direct anterior RT field, which irradiates substantial volumes of cardiac tissue, is now less frequently used. On the other hand, recent studies have demonstrated that some patients receive a radiation dose over 25.00 gray (Gy) to a portion of the heart with contemporary postlumpectomy RT techniques for cancer of the left breast (5, 6), and that these techniques have not reduced the mean radiation dose to the left anterior descending coronary artery (5). A study of electronic portal imaging demonstrated that left breast radiotherapy encompasses the cardiac apex in 9% of cases (7). In addition, some women who receive contemporary RT for cancer of the left breast develop scintigraphic evidence of myocardial perfusion defects within 12 months of RT (8). The use of postlumpectomy RT has increased since the publications of results from clinical trials published in 1981 and 1985 that indicated that most women with early breast cancer could avoid mastectomy by undergoing breast-conserving surgery plus RT (9, 10). The widespread use of postlumpectomy RT indicates the need for study of the association between RT and mortality from MI in the population of women who receive this therapy. An institutional case series compared the risk of fatal MI by laterality among 745 cases with 12-year follow-up and found no difference (11). A similar study of 684 cases in the Stockholm Breast Study Group also found no difference (12). Our objective in this study was to compare the risk of mortality from MI after left-sided postlumpectomy RT to the risk after rightsided postlumpectomy RT in a large unselected breast cancer population. METHODS We conducted a population-based cohort study of cases of female breast cancer newly diagnosed in Ontario between January 1, 1982 and December 31, 1987. Cases of breast cancer have been ascertained by the Ontario Cancer Registry (OCR), whose methods and reliability are described by Robles et al. (13), Clarke et al. (14), and Holowaty and Dale (15). The OCR uses records originally prepared for other purposes (all cancer center registrations, all pathology reports mentioning cancer, all hospital discharge abstracts mentioning cancer, and death certificates). Each case record has a unique numeric identifier. Electronic abstracts of hospital procedures and discharges from the Canadian Institute for Hospital Information (CIHI) were labeled with the unique numeric identifier, and linked to the OCR case records. Only inpatient surgical data are available for this study period. RT records from each cancer center in Ontario were linked to case records by cancer center chart number. The study population consisted of the cases of invasive female breast cancer diagnosed between January 1, 1982

Volume 43, Number 4, 1999

and December 31, 1987, who had a record of lumpectomy as maximal breast surgery within 4 months of diagnosis and a record of postlumpectomy RT within 12 months of diagnosis. We drew the following variables directly from the OCR data for each case: cancer diagnosis, date of diagnosis, age at diagnosis, residence at diagnosis, vital status, survival time, and cause of death. There is no description of cancer stage contained in the registry records. Using the residence at diagnosis information, we assigned each case to a geographic region of Ontario; regions were defined by actual referral patterns to regional cancer centers from local communities (16). We drew the following variables from the procedure data: date of admission for procedure and type of procedure. We defined lumpectomy as any one of the following Canadian Classification of Procedures codes: 97.11, local excision of lesion of breast; 97.27, resection of quadrant of the breast; or 97.28 subtotal mastectomy/partial mastectomy not otherwise specified (17), if the procedure was the maximal surgical procedure on the breast within 4 months after the diagnosis of breast cancer. We used a 4-month cut-off after the date of diagnosis to capture as many breast operations as possible that could have been part of the definitive treatment of the newly diagnosed patient, and to avoid misclassification of mastectomies as lumpectomies. We drew the following variables from the RT records: date of first postlumpectomy RT, code for anatomic region of treatment and laterality, radiation dose, and number of fractions of RT delivered. We labeled RT records as postlumpectomy RT if the patient had a lumpectomy record within 4 months of diagnosis, the first date of postlumpectomy RT within 12 months of diagnosis, and the pertinent anatomic region code, to capture any procedures that could be part of the definitive treatment of the case. Postlumpectomy RT may commence later than 4 months from the date of diagnosis because of delayed wound healing, lengthy waiting lists for RT, and for the administration of complete adjuvant chemotherapy prior to the initiation of RT. We used the laterality descriptor in the RT records as the main study variable. Each record has one of the following laterality descriptors: right, left, bilateral, or unknown. The reliability of this descriptor in the electronic records is unknown. We dichotomized the descriptor by labeling records with the bilateral descriptor (n 5 10) as ‘left’ and by labeling records with the unknown descriptor (n 5 1) as ‘right.’ We used logistic regression to perform a multivariable analysis on the odds of using . 2.00 Gy vs. , 5 2.00 Gy per fraction, on the use of cobalt vs. linac RT, and on the use of boost RT (additional dose of RT to lumpectomy site after completion of whole breast RT). A case was labeled as having had a fatal MI if ICD (International Classification of Diseases) code 410 (MI) was recorded as the cause of death in the OCR. We restricted our study of death codes to ICD 410 (MI) because of the preponderance of this diagnosis in the data from Host et al. (18), Rutqvist and Johansson (3), Boivin et al. (19) and Hancock et al. (20). There are known unreliabilities in the

Fatal MI after postlumpectomy RT

● L. F. PASZAT et al.

757

Table 1. Age and RT characteristics of study population by laterality and event status Left (n 5 1555) Censored cases Age at diagnosis: 20–59 years 60 1 years Breast RT: n Median dose Median number of fractions Cobalt breast RT: n (% of breast RT) Linear accelerator breast RT: n (% of breast RT) Breast boost RT: n (% of breast RT) Median dose Median number of fractions

Fatal MI cases

912 613

3 27

1525 4000 16 1180 (77.4) 345 (22.6) 951 (62.4) 1250 5

Total cases 915 640

30 4000 16 22 (73.3) 8 (26.7) 20 (66.7) 1163 5

reporting of the underlying cause of death (21–24). Cause of death reporting errors in this case are likely nondifferential with respect to the main study variable (laterality of RT). Intercoder variability has been described in the recording of ischemic heart disease codes (25, 26). We compared the time to fatal myocardial infarction, censoring all other causes of death, following postlumpectomy RT between left and right-sided groups using actuarial life table analysis and the log rank test. We compared overall survival, censoring only death due to MI, between left and right groups. We compared survival stratified by age, radiation dose per fraction, and radiation quality (cobalt vs linac). We used Cox proportional hazards regression to model the risk of mortality from MI, adjusting for age at diagnosis, and to model the risk among conditional survival

Right (n 5 1451)

1555 4000 16 1202 (77.3) 353 (22.7) 971 (62.4) 1250 5

Censored cases 855 582 1437 4000 16 1106 (77.0) 331 (23.0) 925 (64.4) 1250 5

Fatal MI cases 4 10

Total cases 859 592

14 4000 16 9 (64.3) 5 (35.7) 9 (64.3) 1250 5

1451 4000 16 1115 (76.8) 336 (23.2) 934 (64.4) 1250 5

cohorts, and strata defined by characteristics of the postlumpectomy RT.

RESULTS 25,570 women were diagnosed with invasive breast cancer (ICD 184) from January 1, 1982 to December 31, 1987. Mean age at diagnosis was 61.2 years. Laterality description from the OCR case records was incomplete. A total of 7394 of the 25,570 women underwent lumpectomy as the maximal surgical procedure on the breast within 4 months of the date of diagnosis. Mean age at diagnosis for women undergoing lumpectomy was 61.1 years. The proportion of cases undergoing lumpectomy increased from

Table 2. Odds ratios (95% confidence interval) for use of radiation quality, fraction size, and boost field Variable Age Region

Year of diagnosis Laterality Radiation quality Fraction size Boost field

Category 20–59 60 1 Ottawa Toronto Hamilton Kingston London NW Ontario Windsor NE Ontario 1982–1983 1984–1985 1986–1987 Left Right Cobalt Linac . 2.00 Gy , 5 2.00 Gy Given Not given

Odds of cobalt vs. linac radiation 1.00 0.62 (0.50, 1.87 (1.36, 1.00 36.0 (13.1, 25.8 (12.7, 0.18 (0.13, 6.47 (2.66, 12.6 (4.38, 3.45 (1.88, 1.00 0.30 (0.20, 0.28 (0.19, 1.13 (0.91, 1.00 – – 4.77 (3.66, 1.00 3.08 (2.37, 1.00

0.76) 2.57) 99.0) 52.4) 0.25) 15.7) 36.5) 6.32) 0.44) 0.40) 1.39)

6.21) 4.00)

Odds of fraction size . 2.00 Gy vs. , 5 2.00 Gy 1.00 1.14 (0.92, 0.46 (0.33, 1.00 3.62 (2.33, 0.04 (0.02, 0.32 (0.23, 0.13 (0.07, 0.41 (0.23, 1.53 (0.86, 1.00 1.49 (1.05, 0.68 (0.50, 0.82 (0.66, 1.00 4.85 (3.75, 1.00

1.42) 0.64) 5.63) 0.07) 0.45) 0.23) 0.72) 2.70) 2.11) 0.94) 1.01) 6.27)

– – 10.5 (8.27, 13.2) 1.00

Odds of boost RT vs. no boost 1.00 0.95 (0.78, 1.14) 0.57 (0.43, 0.77) 1.00 0.08 (0.06, 0.11) 0.62 (0.41, 0.94) 5.89 (4.04, 8.58) 1.14 (0.61), 2.13) 1.13 (0.64, 2.00) 0.98 (0.64, 1.49) 1.00 2.51 (1.89, 3.31) 2.06 (1.58, 2.68) 0.99 (0.82, 1.19) 1.00 3.11 (2.42, 4.00) 1.00 10.9 (8.59, 13.8) 1.00 – –

758


●

Biology

●

Physics


Figure 1: (a) Actuarial probability of death due to MI by treatment laterality, censoring deaths from other causes. (b) Actuarial probability of death due to MI, censoring deaths from other causes. Note that laterality is not known in sub-population that did not receive postlumpectomy RT.

0.17 in 1982 to 0.37 in 1987 (Cochran–Armitage trend test statistic 5 26.2, p 5 0.001). A total of 3006 of the 7,394 women who underwent lumpectomy also underwent postlumpectomy RT. The mean age of women who underwent postlumpectomy RT was 55.7 years. The proportion of lumpectomy cases undergoing postlumpectomy RT increased from 0.29 in 1982 to 0.47 in 1987 (Cochran–Armitage trend test statistic 5 11.1, p 5 0.001). Laterality description of postlumpectomy RT was available for 2995 of 3006 cases. A total of 1555 of the 3006 received left-sided postlumpectomy RT and 1467 received right-sided. These 3006 women form the study population. Median follow-up on this population was 106 months (mean 97 months). Mean radiation dose to the breast was 42.975 Gy divided into a mean number of 18.4 fractions. Mean radiation dose to the breast and mean fraction number were similar between left-sided and right-sided cases (left side 43.138 Gy in 18.7 fractions; right side 42.801 Gy in 18.2 fractions). Median doses and median number of fractions, shown in Table 1, reflect the dose/fractionation practice most commonly used in Ontario during those years (40.00 Gy in 16 fractions, often followed by a boost of 12.50 Gy in 5 fractions). The multivariable analysis (Table 2) showed the odds ratio (OR) for daily fraction size . 2.00 Gy was 0.82 (0.66, 1.01) for cases treated with left-sided postlumpectomy RT compared to right-sided, and demonstrated that cases residing in all regions of Ontario were less likely to receive . 2.00 Gy as the daily fraction size, compared to

the Toronto region, except in the Hamilton region [OR 5 3.62, (2.33, 5.63)]. Cases treated with cobalt were more likely to receive . 2.00 Gy per day (OR 5 4.85, 3.75, 6.27). Of the women, 77.1% received whole breast postlumpectomy RT from cobalt treatment machines; the remainder were treated on a linear accelerator. For cases receiving left-sided RT, the OR for cobalt vs. linac RT was 1.13 (0.91, 1.39). Compared to cases residing in the Toronto region, cases in all regions were more likely to receive cobalt RT compared to linac, except in the London region [OR 5 0.18, (0.13, 0.25)]. The proportion of women who received additional RT to part of the breast (boost) is similar between left-sided and right-sided cases (62.4% left, 64.4% right). For cases treated with left-sided RT, the OR for receiving boost RT was 0.99 (0.82, 1.19). Boosts were more likely with cobalt RT [OR 5 3.11 (2.42, 4.00)] vs. linac, and with . 2.00 Gy per fraction vs. , 5 2.00 Gy [OR 5 10.9 (8.59, 13.8)]. Cases residing in the London region were more likely to receive boost RT compared to those in the Toronto region [OR 5 5.89 (4.04, 8.58)], and those residing in the Ottawa region [OR 5 0.57 (0.43, 0.77)], the Hamilton region [OR 5 0.08 (0.06, 0.11)], and the Kingston region [OR 5 0.62 (0.41, 0.94)] were less likely. With follow-up to December 31, 1995, there were 44 deaths due to MI in the study population. The left-sided group experienced 30 deaths due to MI; the right-sided group 14. The mean age at diagnosis for cases that experienced MI were 71.5 (left side) and 68.1 (right side) com-



759

Figure 2: (a) Actuarial probability of death due to MI by age and treatment laterality, censoring deaths from other causes. (b) Actuarial probability of death due to MI by age, censoring deaths from other causes. Note that laterality is not known in sub-population that did not receive postlumpectomy RT.

pared to 55.7 years for all left-sided and 55.8 years for all right-sided cases. The proportion of cases who received cobalt RT, and the proportion of cases who received ‘boost’ RT to a part of the breast after completing whole breast postlumpectomy RT, were similar between women who did and did not experience a fatal MI, stratified by laterality. Life table analysis, censoring at 14-year follow-up or at death due to any cause other than MI, showed that mortality from MI was more likely in the left-sided group (log rank test p 5 0.02) compared to the right-sided. (Fig. 1). Overall survival, censoring at 14-year follow-up or at death due to any cause other than MI, was similar in the left and right groups (log rank test p 5 0.4, data not shown). Agestratified analysis demonstrated the increased likelihood of mortality from MI among the left-sided group age 60 years and older (p 5 0.01), but no difference between left- and right-sided groups less than 60 years old (p 5 0.68) (Fig. 2). The relative risk (RR) for mortality from MI among women who received left-sided postlumpectomy RT was Table 3. Relative risk of death from MI Variable Laterality Age at diagnosis

Category

n

Right Left 20–59 60 1

1451 1555 1774 1232

* Dose of RT was nonsignificant.

Relative risk (95% confidence interval) 1.00 2.10 (1.11, 3.95) 1.00 8.76 (3.90, 19.68)

2.10 (1.11, 3.95) compared to right-sided, adjusting for age at diagnosis (Table 3). Table 4 presents the RR for fatal MI among overall and age-stratified conditional survival cohorts, with elevated point estimates of RR overall and over age 60 years, but with confidence intervals that include unity. Table 5 presents RR estimates stratified by the RT descriptors: radiation quality, daily RT fraction size, and the use of boost RT. Point estimates of RR among left-sided cases are consistently elevated in radiation quality and fraction size strata, but significantly so only in strata with large numbers of observations (. 2.00 Gy per fraction, cobalt). DISCUSSION We have found a significant association between postlumpectomy RT for cancer of the left breast and mortality from MI. The study variable was the laterality of Table 4. Relative risk of fatal MI in cases treated with left-sided RT among conditional survival cohorts Age cohort All women Women aged 60 1 years

Survival cohorts* 0–5 years . 5 years 0–5 years . 5 years

Relative risk (95% confidence interval) 1.89 (0.81, 4.42) 2.23 (0.86, 5.80) 2.39 (0.93, 6.16) 2.88 (0.93, 8.92)

* Given that a case is alive at the beginning of the interval.

760


●

Biology

●

Physics


Table 5. Relative risk (95% confidence interval) of fatal MI stratified separately by radiation quality, fraction size, and boost field Variable

Cobalt radiation

Laterality left vs right 2.32 (1.07, 5.05) Age 60 1 vs. age 20–59 years 9.07 (3.47, 23.7) Cobalt vs. linac – Frac . 2.00 vs. , 5 2.00 0.55 (0.24, 1.23) Boost vs. no boost 0.97 (0.44, 2.11)

Linear accelerator

Fraction size . 2.00 Gy

Fraction size , 5 2.00 Gy

Boost given

Boost not given

1.50 (0.48, 4.66)

2.60 (1.14, 5.91)

1.27 (0.46, 3.51)

2.09 (0.94, 4.61)

1.75 (0.60, 5.13)

6.11 (1.35, 27.7) –

7.47 (2.81, 19.8) 0.49 (0.21, 1.17)

9.57 (2.16, 42.3) 1.07 (0.36, 3.14)

9.65 (3.33, 27.9) 0.49 (0.22, 1.10)

6.01 (1.69, 21.4) 1.64 (0.43, 6.32)

0.75 (0.22, 2.49) 3.43 (0.84, 14.0)

– 0.84 (0.37, 1.93)

– 2.64 (0.98, 7.15)

0.45 (0.20, 1.02) –

1.00 (0.35, 2.90) –

postlumpectomy RT, which is unlikely to be associated with any treatment selection bias. If physicians had prescribed post-lumpectomy RT less frequently to any women with cancer of the left breast, this would bias against finding an association. There are no plausible mechanisms whereby laterality could bias toward finding an association when none truly exists. We have not compared cases who received RT to cases who did not receive RT because of treatment selection bias, and because of incomplete laterality description among unirradiated cases. This association is consistent with laboratory observations and with observations of outcomes following exposure of cardiac tissue to RT for other diseases. Experiments of cardiac irradiation in dogs have demonstrated severe atherosclerosis in coronary arteries (27, 28). Data from dog and rat models suggest increased cardiac damage associated with increasing fraction size (29 –32). Autopsies have shown severe atherosclerosis around the ostia and in the proximal course of the coronary arteries of young adults treated by RT for Hodgkin’s disease, with findings confined to the irradiated volume (33). This association is also analogous to the associations of adjuvant RT with mortality from all causes other than breast cancer (1), from cardiac diseases (2), and from MI (3), all of which have been observed in populations where the majority of cases were treated with postmastectomy RT prior to the lumpectomy era. This is also analogous to our findings among women with left-sided breast cancer treated with adjuvant RT in the SEER population 1973–1992 (4). This association suggests that irradiation of cardiac tissue during postlumpectomy RT for cancer of the left breast in some women might lead to myocardial infarction or ischemic heart disease in general. The absence of a significant association in institutional series of selected cases with fewer observations (11, 12), or in a clinical trial, does not imply that there is no association to be found. In our study in the SEER population (4), stratified analyses and analyses of conditional survival cohorts indicated that the risk of fatal MI was highest 10 to 15 years after diagnosis before age 60 years, among women with leftsided breast cancer treated by adjuvant RT. This Ontario study does not address the long latency issue for risk among women less than 60 years old, because sufficient follow-up data has yet to be accumulated for that purpose.

This Ontario study has demonstrated an observation of increased risk of fatal MI following left-sided postlumpectomy RT among women aged $ 60 years, that was not observed in our SEER study (4). There are extreme differences in the amount of information about RT between the SEER data and the Ontario data. The SEER data simply contain a dichotomous yes/no variable indicating that RT was administered within 4 months of the diagnosis of the cancer, or within 4 months of a surgical procedure performed on the newly diagnosed cancer. SEER has no description of the anatomic region, dose, fractions, radiation energy, or RT technique, and there are no data on any RT that was administered during the months past the 4-month time rule. The exposure of postlumpectomy cases to RT is much more extensively characterized among the Ontario cases, which allowed this Ontario study design and population to be defined by the study question to a greater degree than in our SEER study. Our study accessed information about postlumpectomy RT during the first year after diagnosis, which gave a fuller description of the use of postlumpectomy RT for newly diagnosed breast cancer. The proportion of lumpectomy cases receiving postlumpectomy RT in Ontario from 1982–1987 relates to the fact that this study of necessity included all cases of breast cancer regardless of: 1. Stage because of the absence of stage data; 2. coexisting illnesses of any severity; and 3. age at diagnosis. In addition, several hundred women were enrolled in a randomized clinical trial in which one study arm excluded postlumpectomy RT. We studied the laterality of postlumpectomy RT, the radiation quality, the dose per fraction, and the use of additional boost doses of postlumpectomy RT; however, only laterality of RT was significantly associated with increased risk. Unfortunately we had no data in this Ontario study or in our SEER study with which to adjust for the volume of cardiac tissue irradiated or the dose to which cardiac tissue (rather than the breast) was exposed. Any changes in RT practice during the study period that might reduce the dose and volume of heart exposed would bias against finding an association. The position of the heart with respect to the radiation fields encompassing the left breast will vary from woman to woman, especially in relation to aspects of body habitus, such as anterior-posterior diameter and the relative location of the heart along that axis, the


relative location of the breast while an individual patient is in the treatment position, and the size of the breast. A case-control study design would be appropriate for further investigation of this problem. CONCLUSIONS Left-sided postlumpectomy RT is associated with an increased risk of fatal MI among cases of breast cancer


761

diagnosed in Ontario between 1982 and 1987, compared to right-sided postlumpectomy RT, apparently in women aged $ 60 years at the time of diagnosis. This association suggests that there may be a risk of fatal MI due to irradiation of cardiac tissue during left-sided postlumpectomy RT, and that it is now important to identify the characteristics of women and of RT practice that are associated with substantially increased risk of mortality from MI, so that alternative options may be chosen.

REFERENCES 1. Early Breast Cancer Trialist’s Collaborative Group. Effects of radiotherapy and surgery in early breast cancer. N Engl J Med 1995;333:1444 –1455. 2. Cuzick J, Stewart H, Rutqvist L, Houghton J, Edwards R, Redmond C, Peto R, Baum M, Fisher B, Host H, Lythgoe J, Ribeiro G, Scheurlen H. Cause-specific mortality in long-term survivors of breast cancer who participated in trials of radiotherapy. J Clin Oncol 1994;12:447– 453. 3. Rutqvist LE, Johansson H. Mortality by laterality of the primary tumour among 55,000 breast cancer patients from the Swedish Cancer Registry. Br J Cancer 1990;61:866 – 868. 4. Paszat LF, Mackillop WJ, Groome PA, Boyd C, Schulze K, Holowaty E. Mortality from myocardial infarction following adjuvant radiotherapy for breast cancer in the SEER cancer registries. J Clin Oncol 1998;16:2625–2631. 5. Fuller SA, Haybittle JL, Smith REA, Dobbs HJ. Cardiac doses in post-operative breast irradiation. Radiother Oncol 1992;25: 19 –24. 6. Gyenes G, Gagliardi G, Lax I, Fornander T, Rutqvist LE. Evaluation of irradiated heart volumes in stage I breast cancer patients treated with postoperative adjuvant radiotherapy. J Clin Oncol 1997;15:1348 –1353. 7. Magee B, Coyle C, Kirby MC, Kane B, Williams PC. Use of electronic portal imaging to assess cardiac irradiation in breast radiotherapy. Clin Oncol 1997;9:259 –261. 8. Gyenes G, Fornander T, Carlens P, Glas U, Rutqvist LE. Myocardial damage in breast cancer patients treated with adjuvant radiotherapy: A prospective study. Int J Radiat Oncol Biol Phys 1996;36:899 –905. 9. Fisher B, Bauer M, Margolese R, Poisson R, Pilch Y, Redmond C, Fisher E, Wolmark N, Deutsch M, Montague E, Saffer A, Wickerham L, Lerner H, Glass A, Shibata H, Deckers P, Ketcham A, Oishi R, Russell I. Five-year results of a randomized clinical trial comparing total mastectomy and segmental mastectomy with or without radiation in the treatment of breast cancer. N Engl J Med 1985;312:665– 673. 10. Veronesi U, Saccozzi R, del Vecchio M, Banfi A, Clemente C, de Lena M, Gallus G, Greco M, Luini A, Marubini E, Muscolino G, Rilke F, Salvadori B, Zecchini A, Zucali R. Comparing radical mastectomy with quadrantectomy, axillary dissection and radiotherapy in patients with small cancers of the breast. N Engl J Med 1981;305:6 –11. 11. Nixon AJ, Manola J, Gelman R, Bornstein B, Abner A, Hetelekidis S, Recht A, Harris JR. No long-term increase in cardiac-related mortality after breast-conserving surgery and radiation therapy using modern techniques. J Clin Oncol 1998;16:1374 –1379. 12. Rutqvist LE, Liedberg A, Hammar N, Dalberg K. Myocardial infarction among women with early-stage breast cancer treated with conservative surgery and breast irradiation. Int J Radiat Oncol Biol Phys 1998;40:359 –363. 13. Robles SC, Marrett LD, Clarke EA, Risch HA. An application of capture-recapture methods to the estimation of the com-

14.

15. 16.

17. 18. 19. 20. 21. 22. 23.

24. 25. 26. 27.

28. 29. 30.

pleteness of cancer registration. J Clin Epidemiol 1988;41: 495–501. Clarke EA, Marret LD, Kreiger N. Cancer registration in Ontario: a computer approach. In: Jensen OM, Parkin DM, Maclennan R, Muir CS, Skeet RG, editor. Cancer registration: principles and methods. Lyon, France: International Agency for Research on Cancer; 1991. p. 246 –257. Holowaty EJ, Dale D. The hospital only project. Health Reports 1993;5:91–95. Mackillop WJ, Groome PA, Zhang-Solomons J, Zhou Y, Feldman-Stewart D, Paszat L, Dixon P, Holowaty EJ, Cummings BJ. Does a centralized radiotherapy system provide adequate access to care? J Clin Oncol 1997;15:1261–1271. Statistics Canada, Health Division, Nosology Reference Section. Canadian classification of diagnostic, therapeutic and surgical procedures. Ottawa: Statistics Canada; 1986. Host H, Brennhovd IO, Loeb M. Postoperative radiotherapy in breast cancer—long-term results from the Oslo study. Int J Radiat Oncol Biol Phys 1986;12:727–732. Boivin J-F, Hutchison GB, Lubin JH, Mauch P. Coronary artery disease mortality in patients treated for Hodgkin’s disease. Cancer 1992;69:1241–1247. Hancock SL, Tucker MA, Hoppe RT. Factors affecting late mortality from heart disease after treatment of Hodgkin’s disease. JAMA 1993;270:1949 –1955. de Faire U, Friberg L, Lorich U, Lundman T. A validation of cause-of-death certification in 1,156 deaths. Acta Med Scand 1976;200:223–228. Gittelsohn A, Senning J. Studies on the reliability of vital and health records. Am J Public Health 1979;69:680 – 689. Jansson B, Johansson LA, Rosen M, Svanstrom L. National adaptations of the ICD rules for classification—a problem in the evaluation of cause-of-death trends. J Clin Epidemiol 1997;50:367–375. Lindahl BI, Glattre E, Lahti R, Magnusson G, Mosbech J. The WHO principles for registering causes of death. J Clin Epidemiol 1990;43:467– 474. Dobson AJ, Gibberd RW, Leeder SR. Death certification and coding for ischemic heart disease in Australia. Am J Epidemiol 1983;117:397– 405. Sorlie PD, Gold EB. The effect of physician terminology preference on coronary heart disease mortality. Am J Public Health 1987;77:148 –152. Bradley EW, Zook BC, Casarett GW, Rogers CC. Coronary arteriosclerosis and atherosclerosis in fast neutron or photon irradiated dogs. Int J Radiat Oncol Biol Phys 1981;7:1103– 1108. McChesney SL, Gillette EL, Powers BE. Radiation-induced cardiomyopathy in the dog. Radiat Res 1988;113:120 –132. Gillette EL, McChesney SL, Hoopes PJ. Isoeffect curves for radiation-induced cardiomyopathy in the dog. Int J Radiat Oncol Biol Phys 1985;11:2091–2097. Lauk S, Ruth S, Trott KR. The effects of dose-fractionation on

762


●

Biology

●

Physics

radiation-induced heart disease in rats. Radiother Oncol 1987; 8:363–367. 31. Schulz-Hector S. Radiation-induced heart disease: Review of experimental data on dose response and pathogenesis. Int J Radiat Biol 1992;61:149 –160. 32. Wondergem J, Franken NAP, van der Laarse A, Persons


CCM, Van Ravels FJM, Reijnart ICW, Strootman EG. Changes in cardiac performance and sympathetic stimulation during and after fractionated radiotherapy in a rat model. Radiother Oncol 1996;38:33– 40. 33. Brosius FC, Waller BF, Roberts WC. Radiation heart disease. Am J Med 1981;70:519 –530.