Cardiac toxicity associated with anthracycline-containing chemotherapy in older women with breast cancer

Abstract

BACKGROUND:

The purpose of this study was to determine the risk of chemotherapy-associated cardiac toxicity, including cardiac dysrhythmia, cardiomyopathy, congestive heart failure, ischemic heart disease, and conduction disorders among breast cancer patients with up to 16 years of follow-up.

METHODS:

The authors studied 19,478 women aged >65 diagnosed with breast cancer in 1991-1997 from 16 regions in the Surveillance, Epidemiology, and End Results program. Incidence density and cumulative incidence of cardiac toxicities were calculated, and the time-to-event (cardiac toxicity) analysis was conducted by using the Cox hazard regression model.

RESULTS:

The excess cumulative incidence of congestive heart failure in Year 10 among patients receiving anthracycline-containing chemotherapy compared with patients without chemotherapy was 4.7% (31.9% vs 27.2%). After adjusting for patient and tumor characteristics, patients receiving anthracyclines were 25% more likely to have congestive heart failure compared with those without chemotherapy (hazard ratio [HR], 1.25; 95% confidence interval [CI], 1.07-1.46). Those receiving other agents did not have a significantly elevated risk of developing congestive heart failure. The adjusted risk of cardiomyopathy was 2-fold higher in women who received anthracyclines (HR, 1.95; 95% CI, 1.44-2.62) and was 16% higher in those receiving other agents (HR, 1.16; 95% CI, 0.97-1.39) compared with those without chemotherapy. The increased risk for developing congestive heart failure, cardiomyopathy, and cardiac dysrhythmias in association with chemotherapy were similar after adjusting for hypertension and diabetes. The risk of ischemic heart disease and conduction disorders were not significantly different among the 3 groups.

CONCLUSIONS:

Anthracycline-containing chemotherapy regimens were associated with an increased risk of congestive heart failure, cardiomyopathy, and cardiac dysrhythmias, but not significantly associated with ischemic heart disease or conduction disorders. Cancer 2009. © 2009 American Cancer Society.

The frequency and severity of chemotherapy-associated toxicities are generally the secondary outcome measures of randomized clinical trials, after the primary endpoint on the efficacy of chemotherapy agents. For example, anthracycline-based combination chemotherapy has been among the most active agents in treating breast cancer and has been beneficial in improving the disease-free and overall survival over the past 2 decades.1 Such trials prospectively collect the most comprehensive and systematic data on adverse events in a monitored setting. In fact, the toxicity profiles of various drugs are often established under such “ideal world” conditions in premarketing clinical trials before approval by the US Food and Drug Administration (FDA).2 This process is particularly true for reporting side effects associated with administration of systemic chemotherapy.3-15 However, clinical trials generally have rigid eligibility criteria, often excluding elderly patients, pregnant women, patients with multiple coexisting diseases, and those taking medications suspected of interacting with the study drug.16, 17 Therefore, study participants (motivated volunteers) in clinical trials often represent a much more homogeneous patient population in terms of gender, age, race/ethnicity, and comorbidity, and they are not representative of the population at large, particularly the elderly population who are often substantially underrepresented in clinical trials.16 Furthermore, clinical trials may not have adequate statistical power to detect rare events or have sufficient follow-up to identify long-term toxicities. Hence, the current system of postmarketing surveillance of long-term toxicities requires substantial improvement.17-30 One approach to obtaining more complete information on drug toxicity is to supplement data from premarketing clinical trials with data from postmarketing drug surveillance to produce better estimates. However, there is a paucity of information on cancer chemotherapy-related toxicities from population-based studies.31, 32

Several reports have been published on chemotherapy-associated cardiac complications, particularly anthracycline-containing agents that are known to be associated with a range of cardiotoxic syndromes.11-13 Cardiotoxicity may occur either immediately at the time of drug administration or months to years after treatment. Two studies recently examined the incidence of congestive heart failure associated with chemotherapy for breast cancer and found a 14% to 35% increase in risks for those receiving chemotherapy.31, 32 The objectives of the current analysis were to extend the scope of these studies: 1) to determine the anthracycline-related risk of a wider range of cardiac disorders, including cardiac dysrhythmia, cardiomyopathy, congestive heart failure, ischemic heart disease, and conduction disorders; 2) to determine a more accurate estimate of chemotherapy-associated cardiac toxicity by restricting the study cohort to those without preexisting cardiac disease; and 3) to examine patients with long-term follow-up extending up to 16 years to identify late cardiac toxicities.

MATERIALS AND METHODS

Data Sources

This study used the Surveillance, Epidemiology and End Results (SEER)-Medicare linked database for Medicare beneficiaries diagnosed with breast cancer. The SEER program, supported by the National Cancer Institute, includes population-based tumor registries in selected geographic areas in 1991-1997: San Francisco/Oakland, Detroit, Seattle, Atlanta, Rural Georgia, Los Angeles county, the San Jose–Monterey area; and the states of Connecticut, Iowa, New Mexico, Utah, and Hawaii, covering >14% of the US population.32 The Medicare program provides payments for hospital, physician and outpatient medical services for >97% of persons aged ≥65. The University of Texas Health Science Center at Houston Committee for the Protection of Human Subjects approved this study.

Study Population

The study population comprised 32,004 women diagnosed with American Joint Committee on Cancer (AJCC) stages I-IV breast cancer as the only primary tumor at age ≥65 years in 1991-1997, who had full coverage of Medicare Parts A and B, and were not enrolled with Health Maintenance Organizations from the year of diagnosis to the last follow-up (December 2005 or date of death). We studied patients from only 1991-1997 to examine the long-term risk of cardiac toxicity associated with anthracycline-based chemotherapy and to minimize the potential impact or confounding of trastuzumab that was a new agent since 1998, which was reported to be associated with an increased risk of cardiac toxicity. For the purpose of this study in determining the relation between chemotherapy and cardiotoxicities, we included patients who were free of cardiovascular diseases at the time of cancer diagnosis, specifically congestive heart failure, cardiomyopathy, cardiac dysrhythmias, conduction disorders, and ischemic heart disease. By doing so, we excluded 11,489 cases with preexisting cardiac diseases, 808 cases who received first chemotherapy after 12 months of diagnosis, and 229 cases with mitoxanthrone, leaving 19,478 subjects for the final analysis.

Variables

Outcome Variables on Cardiac-Toxicities

Patients were defined to have been diagnosed with cardiac diseases when there were at least 2 claims from the ICD-9-CM diagnosis codes34 with >30 days apart after the date of breast cancer diagnosis: congestive heart failure (diagnosis code, 428.x), cardiomyopathy (diagnosis code, 425.x), cardiac dysrhythmia (diagnosis code, 427.x), conduction disorder (diagnosis code, 426.x), and ischemic heart disease (diagnosis codes, 410.x, 411.x, 412.x, 413.x, or 414.x). The time in days was calculated from the date of diagnosis to the date of the first claim for cardiac condition.

Chemotherapy and Other Treatments

Patients were defined as having received chemotherapy if any of the following procedure codes for chemotherapy were present in the inpatient, physician, or outpatient claims within 12 months of breast cancer diagnosis35: the ICD-9-CM procedure code of 9925 and V codes of V58.1, V66.2, or V67.2; the Common Procedure Terminology codes of 96,400-96,549, J8510, J8520, J8521, J8530-J8999, J9000-J9999 and Q0083-Q0085; and revenue center codes of 0331, 0332, and 0335. Of these patients, those who received anthracycline-containing regimens were defined when there were claims for doxorubicin (J9000, J9001, or J9010). A small number of 229 subjects with mitoxanthrone (J9293) were not included in the study cohort. Thus, patients were classified into 3 mutually exclusive groups: 1) anthracycline-containing chemotherapy (n = 1104); 2) nonanthracycline chemotherapy (n = 3398); or 3) no-chemotherapy (n = 14,986). Breast-conserving surgery, or mastectomy, or radiation therapy36 were defined previously.

Socioeconomic Status

The percentage of persons living below the poverty line at the census tract level from the 1990 census for cases between 1991 and 1997 was used to define the socioeconomic status (SES). These percentages were classified into quartiles: ≤3.62%, 3.63%-6.62%, 6.63%-11.99%, and ≥12.00% (poorest SES).

Comorbidity Index

Comorbidity was ascertained from Medicare claims by identifying comorbid conditions between 1 year before and 1 month after the diagnosis of breast cancer, including diabetes, liver disease, and dementia, but excluding any malignant tumors. Detailed methods for creating a weighted comorbidity score have been previously reported.37

Analysis

Differences in the distribution of baseline characteristics among the 3 chemotherapy groups were tested using the chi-square statistic. Incidence rate (density) was defined as the ratio of the number of new cardiac disorders over the total person-years. Person-years were calculated as the number of patients multiplied by the number of years from diagnosis to the date of the first cardiac toxicity, date of death, or date of last follow-up, whichever occurred first. The cumulative incidence (probability) of cardiotoxicity was calculated using the statistical program by Penman and Johnson.38 The Cox hazard regression model was used to perform the time-to-event (toxicity) analysis by using the PHREG procedure available in SAS (Cary, NC: SAS System).

RESULTS

Table 1 presents the distribution of baseline characteristics among the 3 groups of patients stratified by chemotherapy status. Of the 19,478 patients with breast cancer at ages 65-89 years in 1991-1997, 5.7% received anthracycline-based chemotherapy, 17.4% received other chemotherapy agents (of which 45.6% were flurouracil and 9.6% were taxanes), and 76.9% did not receive chemotherapy. A higher proportion of younger or married women received chemotherapy; 50.5% of patients aged 65-69 years received anthracycline-based chemotherapy, 36.3% received other chemotherapy, and 23.9% received no chemotherapy. A slightly lower percentage of Caucasians received anthracycline-based chemotherapy. There were no significant differences in distribution across the quartiles of SES. A higher percentage of patients with earlier stage disease, small tumor size, fewer positive lymph nodes, well-differentiated tumors, and positive hormone receptor status did not receive chemotherapy. Patients with no comorbidities were more likely to receive chemotherapy. A higher proportion of women with bilateral breast cancer and bilateral mastectomies and those in recent years received anthracycline-based chemotherapy. There was no claim for epirubicin (J9178) in these patients.

Table 1. Comparisons of Characteristics Among Women With Breast Cancer According to the Receipt of Chemotherapy

	No. (Column %) of Cases
	No Chemotherapy (n=14,986)		Other Chemotherapy (n=3388)		Anthracycline-Based Chemotherapy (n=1104)
	No.	%	No.	%	No.	%
Median age (range)	75 (65-89)		72 (65-89)		70 (65-89)
Age, y
65-69	3577	23.9	1229	36.3	558	50.5
70-74	3934	26.3	1062	31.4	338	30.6
75-79	3569	23.8	678	20.0	168	15.2
80-84	2535	16.9	318	9.4	32	2.9
85-89	1371	9.2	101	3.0	8	0.7
Race/Ethnicity
Caucasian	13,261	88.5	3007	88.7	947	85.8
African American	820	5.5	192	5.7	89	8.1
Other	905	6.0	189	5.6	68	6.2
Marital status
Married	6281	41.9	1680	49.6	579	52.5
Unmarried	8206	54.8	1621	47.9	501	45.4
Unknown	499	3.3	87	2.6	24	2.2
Socioeconomic status
First quartile (high)	3781	25.2	882	26.0	309	28.0
Second quartile	3824	25.5	814	24.0	262	23.7
Third quartile	3839	25.6	878	25.9	278	25.2
Fourth quartile (low)	3317	22.2	769	22.7	233	21.0
Missing	225	1.5	45	1.3	22	2.0
Tumor stage
I	8124	54.2	1100	32.5	64	5.8
II	3984	26.6	1398	41.3	469	42.5
III	690	4.6	337	9.9	290	26.3
IV	710	4.7	298	8.8	190	17.2
Unstaged	1478	9.9	255	7.5	91	8.2
Tumor size, cm
<1	3040	20.3	383	11.3	39	3.5
1-1.9	5352	35.7	1033	30.5	171	15.5
2-2.9	2648	17.7	713	21.0	210	19.0
3-3.9	1068	7.1	380	11.2	137	12.4
≥4	1466	9.8	549	16.2	370	33.5
Missing	1412	9.4	330	9.7	177	16.0
No. of positive lymph nodes
0 (negative)	8589	57.3	1436	42.4	126	11.4
1	883	5.9	353	10.4	95	8.6
2-3	598	3.9	316	9.3	113	10.2
4-9	438	2.9	310	9.2	189	17.1
10-51	262	1.8	219	6.5	196	17.7
Missing	4216	28.1	754	22.3	385	34.9
Tumor grade
Well differentiated	2182	14.6	276	8.2	41	3.7
Moderately differentiated	4983	33.3	1035	30.6	260	23.6
Poorly differentiated	3505	23.4	1221	36.0	546	49.5
Unknown/Missing	4316	28.8	856	25.3	257	23.3
Hormone receptor status
Positive	9913	69.2	2069	61.1	544	49.3
Negative	1329	8.9	684	20.2	310	28.1
Unknown	3744	24.9	635	18.7	250	22.6
Comorbidity scores
0	11,517	76.9	2725	80.4	916	83.0
1	2622	17.5	527	15.6	155	14.0
≥2	847	5.7	136	4.0	33	2.9
Primary therapy
No cancer-directed surgery	790	5.3	159	4.7	122	11.1
Breast conserving surgery	6103	40.7	1129	33.3	255	23.1
Mastectomy	8093	54.0	2100	62.0	727	65.9
Radiotherapy to chest
Right-side	9258	61.8	1786	52.7	404	36.6
Left-side	2792	18.6	768	22.7	328	29.7
Both sides or unknown	2910	19.4	820	24.2	365	33.1
No radiotherapy	26	0.2	14	0.4	7	0.6
Year of diagnosis
1991	2393	16.0	517	15.3	103	9.3
1992	2173	14.5	529	15.6	126	11.4
1993	2114	14.1	465	13.7	105	9.5
1994	2055	13.7	460	13.6	146	13.2
1995	2080	13.9	472	13.9	168	15.2
1996	2033	13.6	424	12.5	196	17.8
1997	2138	14.3	521	15.4	260	23.6
Total	14,986	100.0	3388	100.0	1104	100.0

Table 2 presents the incidence rate (density) of cardiac toxicities stratified by chemotherapy status and age. For example, the incidence of congestive heart failure in patients aged 65-69 years who received anthracycline-containing chemotherapy was 33.2 per 10,000 person years, compared with 22.3 for those receiving nonanthracycline chemotherapy and 18.5 for those not receiving chemotherapy. Overall, incidence rates increased with advanced age for all 5 cardiac conditions. Incidence rates of congestive heart failure, cardiomyopathy, and cardiac dysrhythmia were generally the highest in patients receiving anthracycline-containing chemotherapy, and the lowest in those without chemotherapy except among the oldest old patients (aged 85-89 years). This is in contrast to the incidence rates of ischemic heart disease and conduction disorders, which did not strongly correlate with chemotherapy status.

Table 2. Incidence Density of Cardiac Toxicity by Chemotherapy Status and Age

Age, y	Incidence Density of Cardiac Toxicity (No. of Cases With Cardiac Conditions Per 10,000 Person-Years)
	No Chemotherapy	Nonanthracycline Chemotherapy	Anthracycline-Containing Chemotherapy
Congestive heart failure
65-69	18.5	22.3	33.2
70-74	27.5	32.3	37.7
75-79	40.8	47.3	66.2
80-84	61.6	58.8	97.8
85-89	83.6	75.8	71.5
Cardiomyopathy
65-69	4.3	5.8	11.3
70-74	4.6	5.5	13.0
75-79	5.5	7.0	19.9
80-84	5.8	8.5	13.4
85-89	6.4	3.7	82.4
Cardiac dysrhythmia
65-69	32.9	39.6	45.5
70-74	42.0	45.2	61.3
75-79	52.0	59.3	61.4
80-84	63.2	59.5	128.5
85-89	80.2	72.5	130.9
Ischemic heart disease
65-69	33.7	37.7	34.9
70-74	40.6	43.3	44.3
75-79	49.5	52.8	59.5
80-84	60.5	58.7	129.8
85-89	71.3	59.8	35.7
Conduction disorder
65-69	5.9	6.1	6.4
70-74	8.5	9.2	13.8
75-79	11.0	10.7	9.1
80-84	14.5	12.2	0.0
85-89	18.3	15.2	0.0

Table 3 presents the cumulative incidence (probability) of congestive heart failure over the 15-year period and was stratified into 3 groups according to chemotherapy status. The incidence probability was 3.1% in Year 1 for anthracycline-containing chemotherapy and 1.5% for other chemotherapy compared with 1.6% for the no chemotherapy group. Cumulative incidence in Year 10 was 31.9%, 26.4%, and 27.2%, respectively, for the above-mentioned 3 groups, indicating that the excess incidence was 4.7% among patients receiving anthracycline-containing chemotherapy, and there was no excess incidence in patients with nonanthracycline-based chemotherapy compared with patients without chemotherapy. Cumulative incidence for congestive heart failure and other 4 cardiac conditions was presented in Figure 1. For example, cumulative incidence curves for congestive heart failure and cardiomyopathy were notably different among the 3 groups with the highest incidence rate in patients receiving anthracyclines and lowest in those without chemotherapy, whereas the incidence curves were not much different for other cardiac conditions among the 3 groups. By Year 10 of follow-up, the number at risk for congestive heart failure was 246 (22.2% of the baseline) for anthracyclines group, 1261 (37.2%) for other chemotherapy group, and 6094 (40.7) for no-chemotherapy group. The incidence at Year 10 was generally low at 6.9%, 6.5%, and 6.2%, respectively, for the 3 groups.

Figure 1. Cumulative incidence of (A) congestive heart failure, (B) cardiomyopathy, (C) cardiac dysrhythmia, (D) ischemic heart disease, and (E) conduction disorder by chemotherapy status are shown.

Table 3. Cumulative Incidence Probabilities of Having Congestive Heart Failure by Chemotherapy Use

Follow-Up Time, y	No. at Risk in Each Time Interval	No. With Congestive Heart Failure	Cumulative Incidence Probability	95% CI
CI indicates confidence interval. * Reported to avoid the cell size of <5 cases as required by the data user agreement.
Anthracycline-containing chemotherapy
0	1104	0	0.000	—
1	961	33	0.031	0.022-0.043
2	771	23	0.057	0.044-0.073
3	675	20	0.083	0.066-0.103
4	586	25	0.118	0.098-0.142
5	517	25	0.157	0.133-0.185
6	474	15	0.183	0.156-0.213
7	441	13	0.205	0.177-0.237
8	411	13	0.229	0.200-0.263
9	350	26	0.279	0.246-0.316
10	246	17	0.319	0.283-0.359
11	167	7	0.342	0.304-0.384
12	111	13	0.400	0.354-0.449
13	70	8	0.454	0.399-0.512
14	36	8	0.529	0.461-0.600
≥15	11	<5*	0.594	0.511-0.679
Nonanthracycline chemotherapy
0	3388	0	0.000	—
1	3134	50	0.015	0.012-0.020
2	2879	52	0.032	0.026-0.039
3	2668	53	0.050	0.043-0.059
4	2479	58	0.072	0.063-0.081
5	2317	66	0.097	0.086-0.108
6	2176	55	0.119	0.107-0.131
7	2032	65	0.145	0.133-0.159
8	1901	70	0.175	0.161-0.190
9	1665	105	0.222	0.206-0.239
10	1261	81	0.264	0.247-0.282
11	959	73	0.311	0.292-0.331
12	690	58	0.358	0.336-0.380
13	482	44	0.405	0.381-0.429
14	276	40	0.465	0.438-0.494
≥15	100	21	0.518	0.485-0.552
No chemotherapy
0	14,986	0	0.000	—
1	13,774	238	0.016	0.015-0.019
2	12,958	219	0.030	0.030-0.036
3	12,163	277	0.054	0.050-0.058
4	11,443	273	0.075	0.071-0.080
5	10,760	311	0.101	0.096-0.106
6	10,111	322	0.128	0.123-0.134
7	9449	345	0.158	0.152-0.165
8	8788	362	0.191	0.184-0.198
9	7879	430	0.232	0.224-0.239
10	6094	377	0.272	0.264-0.280
11	4573	354	0.319	0.310-0.328
12	3252	296	0.369	0.359-0.379
13	2203	210	0.416	0.405-0.427
14	1230	195	0.479	0.466-0.492
≥15	472	99	0.535	0.519-0.551

Table 4 presents the multivariate time-to-event (toxicity) analysis for the relative risk (hazard ratio) of developing various types of cardiac conditions. Patients treated with anthracycline-containing chemotherapy were 25% significantly more likely to develop congestive heart failure compared with those without chemotherapy, after adjusting for patient and tumor characteristics (HR, 1.25; 95% CI,1.07-1.46). Those receiving other chemotherapy did not have a significantly elevated risk of developing congestive heart failure. The adjusted risk of developing cardiomyopathy was twice as likely in women with anthracycline-containing chemotherapy (HR, 1.95; 95% CI, 1.44-2.62) and was 16% higher in those treated with other chemotherapy (HR, 1.16; 95% CI, 0.9710-1.39) compared with those without chemotherapy. The risk of cardiac dysrhythmia was also slightly but significantly elevated in women with anthracycline-containing chemotherapy (HR, 1.16; 95% CI, 1.02-1.33) compared with those without chemotherapy. However, the risk of ischemic heart disease and conduction disorders was not significantly different in women with anthracycline-containing chemotherapy and in those who received other chemotherapy compared with those without chemotherapy. The risk of cardiac toxicities remained similar after adjusting for several known preexisting risk factors for cardiovascular disease such as hypertension and diabetes. For example, after adjusting for both hypertension and diabetes, the hazard ratio in patients receiving anthracyclines compared with no chemotherapy was 1.27 (95% CI, 1.09-1.49) for congestive heart failure, 1.98 (95% CI, 1.47-2.67) for cardiomyopathy, 1.17 (95% CI, 1.02-1.34) for cardiac dysrhythmia, 1.00 (95% CI, 0.87-1.16) for ischemic heart disease, and 1.01 (95% CI, 0.74-1.39) for conduction disorders. These hazard ratios were similar to those summarized in Table 4 without adjusting for hypertension and diabetes.

Table 4. Hazard Ratio of Cardiac Toxicity in Patients Receiving Anthracycline Chemotherapy

Patient and Tumor Characteristics	Hazard Ratio* (95% CI) of Having Cardiac Toxicity
	Congestive Heart Failure	Cardiomyopathy	Cardiac Dysrhythmia	Ischemic Heart Disease	Conduction Disorder
CI indicates confidence interval. * Hazard ratio was adjusted for variables listed in the table, plus ethnicity, marital status, socioeconomic status, tumor stage, size, grade, number of positive lymph nodes, hormone receptor status, surgery, and year of diagnosis.
Chemotherapy
No chemotherapy	1.00 (ref)	1.00 (ref)	1.00 (ref)	1.00 (ref)	1.00 (ref)
Other chemotherapy	1.03 (0.88-1.11)	1.16 (0.97-1.39)	1.06 (1.00-1.14)	1.05 (0.99-1.13)	0.97 (0.83-1.12)
Anthracycline-containing chemotherapy	1.25 (1.07-1.46)	1.95 (1.44-2.62)	1.16 (1.02-1.33)	0.99 (0.86-1.15)	1.00 (0.73-1.37)
Age, y
65-69	1.00	1.00	1.00	1.00	1.00
70-74	1.47 (1.35-1.60)	1.06 (0.89-1.26)	1.29 (1.21-1.38)	1.25 (1.17-1.34)	1.55 (1.33-1.80)
75-79	2.23 (2.06-2.43)	1.34 (1.11-1.62)	1.68 (1.57-1.80)	1.61 (1.50-1.73)	1.98 (1.70-2.31)
80-84	3.38 (3.08-3.70)	1.46 (1.15-1.85)	2.20 (2.02-2.38)	2.13 (1.96-2.32)	2.78 (2.33-3.32)
85-89	4.96 (4.25-5.50)	1.71 (1.23-2.39)	3.16 (2.84-3.52)	2.88 (2.58-3.22)	3.93 (3.13-4.93)
Race/Ethnicity
Caucasians	1.00	1.00	1.00	1.00	1.00
African Americans	1.34 (1.19-1.50)	1.94 (1.51-2.49)	1.31 (1.18-1.46)	1.44 (1.30-1.60)	1.48 (1.18-1.85)
Others	0.76 (0.67-0.87)	0.94 (0.70-1.26)	0.73 (0.66-0.82)	0.87 (0.78-0.97)	0.72 (0.56-0.93)
Comorbidity scores
0	1.00	1.00	1.00	1.00	1.00
1	1.95 (1.82-2.08)	1.66 (1.40-1.96)	1.49 (1.40-1.58)	1.82 (1.71-1.93)	1.48 (1.30-1.69)
≥2	2.65 (2.38-2.95)	2.25 (1.70-2.97)	1.69 (1.51-1.89)	2.48 (2.25-2.75)	1.69 (1.33-2.15)
Radiotherapy to chest
Right side	1.00	1.00	1.00	1.00	1.00
Left side	1.14 (1.04-1.24)	1.00 (0.80-1.24)	1.09 (1.01-1.18)	1.12 (1.03-1.22)	1.02 (0.86-1.22)
Both sides or unknown	1.01 (0.93-1.11)	1.14 (0.93-1.40)	1.01 (0.93-1.08)	1.06 (0.98-1.15)	1.05 (0.89-1.23)
No radiotherapy	1.35 (0.71-2.56)	1.91 (0.45-8.20)	1.20 (0.68-2.12)	1.66 (0.92-2.99)	1.77 (0.55-5.73)

Table 4 also presents the risk of cardiac toxicities in association with other factors including age, comorbidity, and radiotherapy. The risk of all cardiac toxicities increased significantly with age and comorbidity scores. Compared with those patients who received radiation to the right chest, women treated with radiation to the left side were 14% more likely to develop congestive heart failure, 9% more likely to develop cardiac dysrhythmia, and 12% more likely to develop ischemic heart disease, but they had no significant elevated risk of developing cardiomyopathy and conduction disorders.

Table 5 presents the risk of cardiac toxicity in association with the number of chemotherapy claims as a proxy measure for the number of chemotherapy cycles, stratified by tumor stage. In those with stage I-IIIA, patients receiving 5 to 9 doses and ≥10 doses of anthracycline-containing chemotherapy were more likely to develop congestive heart failure compared with those without chemotherapy but with wide confidence intervals due to small numbers after stratification (HR, 1.54; 95% CI, 0.99-2.38, and HR, 1.21; 95% CI, 0.88-1.67, respectively). In those with advanced stage disease (IIIB-IV) or unstaged, patients receiving ≥10 doses of anthracycline-containing chemotherapy were 37% more likely to develop congestive heart failure (HR, 1.37; 95% CI, 0.80-2.34). Patients receiving only 1 to 4 doses of anthracycline-containing chemotherapy or other chemotherapy had slightly elevated but nonsignificant risk of congestive heart failure in patients with late stage. There was also a strong dose-response relation between the receipt of anthracyclines and the risk of developing cardiomyopathy. Higher doses of anthracyclines were also associated with slightly elevated risks of cardiac dysrhythmia and ischemic heart disease, but not with conduction disorders.

Table 5. Hazard Ratio of Cardiac Toxicity in Patients With Breast Cancer by the No. of Chemotherapy Claims and Tumor Stage

No. of Chemotherapy Claims (Doses)	No. of Cases	Hazard Ratio* (95% CI) of Having Cardiac Toxicity
		Congestive Heart Failure	Cardiomyopathy	Cardiac Dysrhythmia	Ischemic Heart Disease	Conduction Disorder
CI indicates confidence interval. * Hazard ratio was adjusted for variables listed in the table, plus ethnicity, marital status, socioeconomic status, tumor stage, size, grade, number of positive lymph nodes, hormone receptor status, surgery, and year of diagnosis.
Tumor stage I-IIIA
No chemotherapy	12,387	1.00 (ref)	1.00 (ref)	1.00 (ref)	1.00 (ref)	1.00 (ref)
Nonanthracycline chemotherapy
1-4	1241	0.99 (0.89-1.11)	0.93 (0.69-1.24)	1.10 (1.00-1.20)	1.06 (0.97-1.17)	1.03 (0.84-1.26)
5-9	494	1.01 (0.84-1.21)	1.52 (1.07-2.17)	0.95 (0.81-1.12)	1.16 (0.99-1.35)	0.96 (0.68-1.35)
≥10	927	1.08 (0.94-1.48)	0.95 (0.67-1.36)	1.06 (0.93-1.20)	0.97 (0.85-1.11)	1.03 (0.78-1.36)
Anthracycline chemotherapy
1-4	412	1.17 (0.93-1.48)	1.47 (0.95-2.26)	1.00 (0.83-1.21)	1.03 (0.85-1.24)	1.07 (0.70-1.62)
5-9	184	1.21 (0.88-1.67)	2.55 (1.56-4.16)	1.31 (1.03-1.67)	1.16 (0.89-1.51)	1.32 (0.77-2.26)
≥10	60	1.54 (0.99-2.38)	2.35 (1.18-4.67)	1.28 (0.91-1.81)	1.28 (0.89-1.84)	0.38 (0.09-1.52)
Tumor stage IIIB-IV or unstaged
No chemotherapy	2599	1.00 (ref)	1.00 (ref)	1.00 (ref)	1.00 (ref)	1.00 (ref)
Nonanthracycline chemotherapy
1-4	347	1.05 (0.83-1.33)	1.10 (0.59-2.04)	1.01 (0.81-1.26)	0.89 (0.69-1.14)	9.92 (0.55-1.54)
5-9	137	0.99 (0.68-1.44)	1.52 (0.63-3.63)	1.26 (0.90-1.76)	0.80 (0.52-1.25)	1.02 (0.46-2.25)
≥10	242	0.96 (0.73-1.26)	0.99 (0.46-2.10)	0.82 (0.61-1.09)	1.01 (0.75-1.36)	0.22 (0.07-0.69)
Anthracycline chemotherapy
1-4	230	0.98 (0.68-1.40)	2.14 (1.14-4.03)	1.11 (0.82-1.51)	0.77 (0.54-1.12)	0.40 (0.16-0.96)
5-9	156	1.25 (0.86-1.83)	3.83 (2.00-7.35)	1.44 (1.04-2.00)	1.29 (0.87-1.91)	1.07 (0.49-2.32)
≥10	62	1.37 (0.80-2.34)	4.68 (2.16-10.14)	1.63 (1.07-2.48)	1.37 (0.83-2.27)	0.69 (0.20-2.34)

DISCUSSION

The potential long-term cardiac risk that results from anthracycline-containing chemotherapy is relatively well-studied in children,39 but it is less well-studied in adult patients with cancer, particularly in the elderly population. Our study sought to examine whether anthracycline-based or nonanthracycline-based chemotherapy is associated with the increased risk of dysrhythmia, ischemic heart disease, and conduction disorders, in addition to congestive heart failure and cardiomyopathy. The overview of the 194 randomized trials of adjuvant chemotherapy with up to 15 years of follow-up by the Early Breast Cancer Trialists' Collaborative Group1 reported that the relative risk of death from heart disease was 1.33 in women receiving anthracycline-based versus nonanthracycline-based chemotherapy. However, this overview did not report the incidence of heart disease and did not differentiate between the various types of heart disease. Our study showed a relative risk (HR, 1.25) for the incidence of congestive heart failure in patients receiving anthracycline-based chemotherapy compared with those without chemotherapy in 1991-1997 with up to 16 years of follow-up. Also, our study showed an excess risk of incidence of 4.7% in patients receiving anthracyclines compared with those without chemotherapy, whereas there was a survival advantage of 4.2% in patients receiving anthracyclines as reported in the above overview of clinical trials.1 This is consistent with the current literature illustrating that anthracyclines prolong survival while carrying a risk of cardiac toxicitiy.40-45

A recent study of 1176 patients from the Southwest Oncology Trial40 concluded that exposure to doxorubicin did not increase the risk of adverse cardiac effects (specifically deterioration in left ventricular ejection fraction). However, although left ventricular ejection fraction is a marker for advanced myocyte damage, approximately half of all heart failure diagnosed in the United States occurs in patients who maintain a normal left ventricular ejection fraction.41 Therefore, the association between anthracycline use and risk of late heart failure could be underestimated if the estimate is based on left ventricular ejection fraction alone. A notable finding from our study is that the risk of cardiac toxicity had a dose-response relation with chemotherapy particularly among those patients who received 10 or more doses of anthracycline-containing chemotherapy. The above-mentioned findings remain similar after controlling for preexisting hypertension and diabetes. Conversely, there were no significant associations between the risk of ischemic heart disease and conduction disorder and the use of anthracycline-containing chemotherapy.

This study possesses several strengths. First, Medicare claims cover all medical services provided, including inpatient hospitalization, outpatient clinics, and physician office visits, regardless of where patients sought medical care in the United States, thus ensuring more comprehensive information on treatment and associated medical conditions. Second, a large retrospective cohort of breast cancer patients had been followed up for up to 16 years, allowing for more adequate assessment of long-term or late toxicities that may be associated with chemotherapy administration. Third, SEER-Medicare data not only provide well-validated pathologic information on tumor characteristics (stage and histology) at diagnosis but also allow for the examination of socioeconomic factors. Furthermore, comorbid conditions can be identified from Medicare data. Because comorbidity is a known confounder for treatment and cardiac conditions, controlling for comorbidity becomes critical in minimizing residual confounding when addressing the cardiac risk associated with chemotherapy.

It is important to note some limitations of this study. First, Medicare claims have limited information on the dose and intensity of chemotherapy administered that could have affected the occurrence and severity of cardiac toxicities. This study relied on common procedure codes that specified standard dose for each chemotherapy agent; but, in practice, physicians may have modified a standard chemotherapy dose for an individual patient according to preexisting medical conditions or tolerability. Furthermore, we used the number of chemotherapy claims as a proxy measure for the number of cycles of chemotherapy administered. Although this study demonstrated an interesting dose-response relationship between number of cycles of chemotherapy and the risk of developing cardiac toxicity, which is consistent with clinical trial reports, the estimates could potentially be diluted because of the lack of “true” cycles and doses that can be obtained in closely monitored trials. Third, although Medicare claims on overall chemotherapy administration have been externally validated, the validity of claims on cardiac disorders has not been well-studied. There may also be potential surveillance bias in ascertaining the cardiac outcomes. For example, patients after receiving anthracyclines may be more likely to see their physicians for check-ups and more likely to be labeled as having congestive heart failure and cardiomyopathy, leading to differential misclassification bias. Conversely, some patients may have had preexisting cardiomyopathy or low left ventricular ejection fraction, but they may not be labeled as suffering from this condition, leading to nondifferential misclassification bias. There may also be potential selection bias when physicians make decisions on the prescription of chemotherapy based on individual's cardiac function or performance status. Although we excluded those with history of cardiovascular diseases, some unmeasured preclinical symptoms that are related with heart disease could potentially lead to selection bias. Also, the number of patients was large at baseline, but after more than 10 years of follow-up, the number of patients became smaller, which could affect the stability of incidence estimates. In addition, if patients had private medical services with claims that were not filed with Medicare for reimbursement, these data might be missed, thus leading to bias toward diluting the exposure-outcomes association. Furthermore, this study did not address physician or hospital characteristics that might be associated with selection of chemotherapy and intensity of surveillance for monitoring patients.

In conclusion, anthracycline-containing chemotherapy was associated with an increased risk of developing congestive heart failure, cardiomyopathy, and cardiac dysrhythmias, but it was not significantly associated with the development of ischemic heart disease and conduction disorders. Other (nonanthracycline) chemotherapy agents were not significantly associated with an increased risk of congestive heart failure and conduction disorders, but they were associated with a slightly elevated risk of cardiomyopathy, dysrhythmias, and ischemic heart disease.

Conflict of Interest Disclosures

This study was supported by a grant from the Agency for Healthcare Research and Quality (R01-HS016743).