|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2003 American Society for Clinical Oncology Diminished Survival in Patients With Inner Versus Outer Quadrant Breast Cancers
From the Department of Radiation Oncology and Huntsman Cancer Institute, University of Utah, Salt Lake City, UT. Address reprint requests to David K. Gaffney, MD, PhD, Department of Radiation Oncology, University of Utah, 50 North Medical Drive, Salt Lake City, UT, 84132; email: david.k.gaffney{at}hsc.utah.edu.
Purpose: To evaluate the role of breast tumor location on survival in patients with breast cancer. Patients and Methods: We evaluated tumor location within the breast on breast cancer–specific survival (BCSS) and overall survival (OS) in patients with invasive breast cancer using the Surveillance, Epidemiology, and End-Results (SEER) registries in the United States. Effects on survival were evaluated according to age, stage, tumor site, tumor size, grade, axillary lymph node status, extent of surgery, and radiotherapy (RT). A multivariate model was used with complete data on 45,880 patients. The median follow-up was 59 months. Results: Patients with outer tumor location demonstrated superior BCSS on Kaplan-Meier analysis for both local stage (node-negative, P < .001) and regional stage disease (node-positive, P = .0002). For BCSS, the hazard ratio (HR) for inner quadrant location compared with outer quadrant was 1.31 (95% confidence interval [CI], 1.19 to 1.37; P < .001); and for OS, the HR was 1.12 (95% CI, 1.05 to 1.17; P < .001). When ER and PR status were included in the model, the HR for inner quadrant location compared with outer quadrant was 1.27 for BCSS (95% CI, 1.16 to 1.40; P < .001) and 1.11 for OS (95% CI, 1.03 to 1.19; P = .004). Patients treated by lumpectomy that received RT had a superior OS compared with patients that did not receive RT in both local (HR, 0.52; 95% CI, 0.48 to 0.61; P < .001) and regional (HR, 0.63; 95% CI, 0.56 to 0.72; P < .001) stage disease. Mastectomy patients with local stage disease that received RT had a diminished OS (HR, 1.24; 95% CI, 1.02 to 1.50; P = .033). Conclusion: On multivariate analysis, incorporating data on age, stage, tumor site, tumor size, grade, ER and PR status, axillary lymph node status, extent of surgery, and RT, this SEER registry–based study demonstrates that medial tumor location adversely impacts BCSS and OS.
RECENT STUDIES have implicated a medial or central breast tumor location as a negative prognostic factor on risk of relapse and overall survival (OS).1–3 In two of the studies, the magnitude of the impact on survival was large, with a 20% to 46% increase in mortality in patients with inner or central breast cancers. Currently, there are many prognostic factors to identify patients at high risk for relapse; however, tumor site is generally not included. A proposed mechanism for the increased risk of metastases and death from breast cancer from medial tumors is occult involvement of internal mammary lymph nodes (IMN) that are not systematically treated with either surgery or radiation therapy (RT). With the advent of sentinel lymph node biopsies, primary drainage to the IMN occurs in 9% to 45% of breast cancers; however, the rates of histologic positivity seems to be lower.4–6 Extended radical mastectomy series indicate that the rate of positive IMN for patients with negative and positive axillary lymph nodes ranges from 8% to 13.7% and 28% to 48%, respectively.7–10 A review of more than 7,000 patients has shown that isolated IMN metastases occurs in approximately 5% of cases.11 It is worthwhile to consider that these rates of IMN positivity are likely an overestimate of what would be observed today because most breast cancer services have observed an increasing percentage of women presenting with smaller tumors compared with two and three decades ago. To discern the impact of tumor location within the breast, we sought to quantify the effect on breast cancer–specific survival (BCSS) and OS of tumor location in a large cancer registry.
We performed a retrospective cohort study evaluating the tumor site within the breast on BCSS and OS in patients with invasive carcinoma using the Surveillance, Epidemiology, and End-Results (SEER) registries in the United States. The national SEER cancer registry consists of nine regional registries and is known to be of high quality with high rates of completeness of case ascertainment. The SEER database includes 409,544 breast cancer records; 235,071 patients were diagnosed between 1988 and 1997. Patients were included who were diagnosed between 1988 and 1997 and for whom complete data were available. The following prognostic factors were included in the analysis. Age and tumor size were treated as continuous variables. Stage, quadrant location, grade, axillary lymph node status, extent of surgery, estrogen receptor (ER) and progesterone receptor (PR) status, and RT were treated as categorical variables. The following were causes for exclusion: patients with distant disease, male breast cancers, cases with missing variables, and patients that did not have pathologic staging of the ipsilateral axilla. When the model included data on ER and PR, 32,940 patients were evaluated; otherwise, complete data were available on 45,880 patients. These patients serve as the study population for this report. All participants in the SEER program routinely link their files with vital records (ie, death certificates) in their respective areas of coverage to identify cancer patients that have died (regardless of cause of death). Thus, death certificates are the source for information regarding underlying cause of death as recorded in the SEER program database. In the United States, death certification is conducted according to standardized procedures promulgated by the National Center for Health Statistics. Underlying cause of death is typically coded by trained nosologists in each state, in accordance with standardized methods and on the basis of information regarding cause of death as recorded by the attending physician. Furthermore, the National Center for Health Statistics conducts routine edits of death certificates to ensure quality of data. Although problems with death certification are well documented, this mechanism still remains the best available in this context. BCSS was used as the primary end point because it is a vital statistic that directly relates to the disease process and can be tracked with high precision in the SEER database. Use of BCSS minimizes confounding of survivorship by other causes of morbidity. No difference in BCSS was found between the outer and central tumor locations as well as between poorly differentiated and undifferentiated tumors, so the respective categories were combined. The effects of categorical variables on the hazard ratio (HR) for BCSS were modeled using dummy variables to code the effect of being in some category as compared with the baseline category. Multivariate proportional hazards regression was used to study the association between the prognostic factors and survival end point. Univariate analysis by the log-rank test as well as backward and forward multivariate variable selection procedures on the basis of the likelihood ratio test were used to find a combination of variables prognostic for the end point. Analysis of partial residuals was used to verify the adequacy of the proportional hazards assumption with respect to the factors entered into the model. The median follow-up was 59 months in the multivariate model. Actuarial calculations according to the Kaplan-Meier method were performed.12 The adequacy of the proportional hazards assumption was tested using a hierarchical nonproportional semiparametric model incorporating the long-term and short-term survival effects.
In the SEER database from 1988 to 1997, we identified 29,802 patients with local (axillary node–negative) disease and 16,078 patients with regional (axillary node–positive) disease. BCSS was strongly influenced by tumor location. A multivariate model was constructed to adjust for differences in other known prognostic factors. For multivariate analysis, all the results obtained refer to effects of particular variables that are adjusted for all other variables in the study. Prognostic factors examined include tumor location, tumor size, number of involved axillary lymph nodes, grade, stage, extent of surgery, and RT. Complete inclusion of all prognostic factors was a prerequisite for inclusion in the multivariate model. Thus, a total of 45,880 patients were included in this analysis. Patient and tumor characteristics are listed in Table 1  . Local stage disease comprised 65.0% of the population. The majority of the patients (78.0%) had their tumor located in the outer or central regions of the breast (Table 1). Patients with inner quadrant lesions were more likely to have local disease (75.6%) compared with patients with outer quadrant lesions (62.4%). Conservative surgery was performed in 43.2% of patients, and mastectomy was performed in 56.8% of patients. RT was performed in 44.4% of the total population and was used infrequently in patients treated with mastectomy (12.2%). The median follow-up period was 59 months, and 4,331 patients (9.4%) died from breast cancer.
The effect of tumor location on BCSS is shown for patients with localized or regional disease in the patient data set used for the multivariate model (Fig 1 ). Patients with outer tumor location demonstrated superior BCSS on Kaplan-Meier analysis for both local stage (node-negative, P < .001) and regional stage (node-positive, P = .0002). For patients with local stage disease, the 5-year BCSS for inner versus outer tumor location was 0.949 ± 0.003 fraction surviving versus 0.960 ± 0.002, and the 5-year OS was 0.897 ± 0.004 versus 0.899 ± 0.002, respectively. Similarly, for patients with regional stage disease, the 5-year BCSS for inner versus outer tumor location was 0.789 ± 0.009 versus 0.822 ± 0.004, and the 5-year OS was 0.728 ± 0.010 versus 0.747 ± 0.004, respectively. On multivariate analysis for BCSS, the HR for inner quadrant location compared with outer quadrant was 1.31 (95% confidence interval [CI], 1.19 to 1.37; P < .001; Table 2), and for OS, the HR was 1.12 (95% CI, 1.05 to 1.17; P < .001). Age at diagnosis was evaluated in the multivariate model and was not found to be of independent prognostic significance as a function of tumor location. Patients treated by lumpectomy that received RT had a superior BCSS and OS compared with patients that did not receive RT in both local and regional stage (Tables 2 and 3). In the initial multivariate model (Table 2), mastectomy patients with local stage that received RT had a diminished BCSS and OS. Only significant effects are shown in Tables 2 and 3. Receipt of RT did not result in improved survival in node-positive mastectomy patients.
To evaluate the effect of ER and PR status in the multivariate model, separate multivariate analyses were performed (Table 3  ). A total of 32,940 patients were available that included the previous prognostic factors and data on ER and PR status. The negative impact of medial quadrant location persisted, with a HR of 1.27 for BCSS (95% CI, 1.16 to 1.40; P < .001) and 1.11 for OS (95% CI, 1.03 to 1.19; P = .004). When hormonal receptor status was included in the model (Table 3), RT did not result in diminished OS for local-stage patients treated by mastectomy.
This SEER registry–based study of breast cancer patients demonstrates that tumor location significantly impacts BCSS in patients with localized and regional disease. These findings are compatible with a growing body of literature examining tumor site as a prognostic factor in breast cancer.1–3 In Zucali et al1 and Lohrisch et al,2 outer quadrant location was observed in 65.7% and 61.5% of patients, respectively, compared with 78.0% with an outer or central primary location in our study. The magnitude of the negative impact of medial tumor location was relatively similar in the three studies. In Zucali et al,1 the HR for distant recurrence was 1.29. In Lohrisch et al,2 the HR was 1.46 for BCSS, and in our study, the HR for BCSS was 1.31. Important differences exist between the three studies. The other studies used patients treated at single institutions. Thus, there was likely greater consistency in patient treatment. In our study, complete data were available on 45,880 patients. The Zucali et al and Lohrisch et al studies included 2,396 patients and 6,781 patients, respectively. A large sample size was required for us to observe significant findings. Initially, the Utah Cancer Registry (one of the nine SEER sites) was used, and no significant findings were identified. The requirement for the large sample may be related to the heterogeneity of patient factors or treatment factors in the nationwide SEER database. For example, the patient population in Zucali et al1 was homogeneous in that all patients were treated by lumpectomy and all patients received 50 Gy to the breast and a 10-Gy boost. The significant impact on BCSS was notable in the other two reports given the much smaller sample size. The other reports were also able to incorporate site of recurrence (local or distant) and time of recurrence. In our study, site of recurrence and time of recurrence were not included to an appropriate extent, although vital status and cause of death are highly accurate. Thus, we chose BCSS and OS as the principle end points. The diminished BCSS in medial lesions observed here and in other studies may be a result of occult spread to IMNs that are not systematically treated. Metastasis to IMNs alone occurs in 5% of breast cancer patients.11 This is similar to the BCSS difference observed in this study and the 5.1% BCSS difference in Lohrisch et al.2 Improved local control with targeting of the internal mammary chain within RT ports may be one way to improve BCSS. This is currently being investigated in a National Cancer Institute of Canada trial in intact breast cancer patients and in a European Organization for Research and Treatment of Cancer (22922) trial of postmastectomy patients. A previous surgical trial showed no difference in survival.10 Similarly, the Oslo and Stockholm trials did not show an improvement in overall survival in patients that received irradiation of regional nodes.13,14 In contrast, one retrospective analysis evaluating treatment to the IMN and three recent randomized trials in axillary node–positive postmastectomy patients revealed a relatively large impact on OS for RT, with an improved HR similar to the magnitude of systemic chemotherapy or tamoxifen.15–20 A modest overall improvement in survival would be expected because isolated metastases to IMNs occur only in approximately 5% of patients. RT is a possible mechanism for a promotion of improved BCSS. The meta-analysis by Peto et al21 showed a favorable impact of RT. In addition, RT has been demonstrated to promote a similar rate of local control to regional lymph nodes compared with surgery.22 Strengths of our study include use of a robust statistical model. We evaluated for both short- and long-term effects on survival, and the adequacy of the proportional hazards modeling was supported. There are several significant limitations of this study. The SEER database is a nationwide registry, and as such, there will be a significant amount of heterogeneity in treatment, such as extent of surgery and use of adjuvant chemotherapy and RT. In addition, this study is a retrospective analysis, and the application of RT was not randomized or balanced. For example, local-stage patients that received RT had a significant larger mean tumor size than patients that did not receive RT (27 v 19 mm, respectively; t test, P < .001). The SEER database does not include details of RT administration, such as whether IMNs were included in breast cancer ports. In standard tangential RT ports, complete and partial inclusion of IMNs occurs in 14% and 40% of patients, respectively.23 It is possible that the benefits of RT reported herein may be at least partially attributable to the effects of systemic therapy and/or hormonal therapy, if such modalities were disproportionately associated with RT. Incomplete information for chemotherapy and hormonal therapy in the central cancer registry limited our ability to appropriately incorporate these factors in this analysis. The sensitivity for the receipt of adjuvant systemic therapy in two registry-based studies were 27% and 55.6%.24,25 Thus, the effect of RT as it relates to tumor location in this report is only intended for generating hypotheses and to encourage enrollment onto prospective studies designed to address the role of RT to the IMN chain. Information on cause of death for deceased breast cancer patients is remarkably complete in the SEER program database. For example, in the SEER program database, out of 44,192 breast cancer deaths in patients diagnosed from 1988 to 1997, a valid cause of death code was identified for 42,501 patients (96.2%). Of the deceased subjects that were included in the present analysis, 95.9% had a valid cause of death. The percentage of patients that had a valid cause of death did not vary significantly between those whose tumors were located in the inner quadrant (95.3%) and those whose tumors were located in the central portion or outer quadrant of the breast (96.0%, P = .1118). There is evidence that breast cancer is accurately coded on death certificates of women who die of this disease. Percy et al26 examined death certificates of deceased women whose only known cancer diagnosis was that of breast cancer. Of those women with breast cancer who subsequently died and had a cancer-related cause of death, 95.0% were correctly coded as having died of breast cancer. We are unaware of compelling evidence to suggest that death certification would have varied between the comparison groups in this investigation. For these reasons, we believe that results from this study were not significantly influenced by potential biases in death certification. Most central cancer registries that participate in the SEER program routinely visit RT units within their areas of coverage to ascertain diagnostic and treatment information for cancer patients that are seen at such facilities. Nonetheless, results from one study indicated that RT was not documented by the SEER program in approximately 18% of female breast cancer patients who were 65 years or older (on the basis of treatment data as recorded on Medicare reimbursement forms).27 Results from this single study suggest that it may be appropriate to interpret information regarding RT as recorded in the SEER program with caution. Nonetheless, if ascertainment of information regarding RT was unrelated to factors under investigation, we would anticipate that underreporting of RT was randomly distributed between the comparison groups. We would further anticipate that random misclassification of RT would bias the result toward the null value (ie, no difference between the comparison groups). Unfortunately, we have no mechanism for evaluating the validity of the RT data among subjects in this investigation. Because this is a retrospective study, caution is urged in evaluating subgroups. For mastectomy patients with local stage disease, RT resulted in an increased HR (1.24; 95% CI, 1.02 to 1.50) for death in the multivariate model; however, this was no longer significant when ER and PR status were included in the model. The marked increase for this subgroup in the HR on BCSS (1.83; 95% CI, 1.42 to 2.53) for receipt of RT suggests that there may be unidentified prognostic factors involved or imbalance between the two groups. Prospective, randomized data in premenopausal patients with high-risk, node-negative breast cancer have been shown to benefit from RT. In the Danish 82b trial, there was a 12% improvement in survival at 10 years in node-negative patients that received RT.16 Significant controversy surrounds targeting of the IMN with RT.28 Recent analysis of patients treated in the Danish randomized trials that received RT to the chest wall and regional lymphatics including the IMN did not show an adverse effect on cardiovascular events.29 In addition, a review of breast-conserving therapy in 745 patients with a minimal potential follow-up of 12 years did not show any deleterious effects of left-sided irradiation with modern RT techniques.30 However, the overview analysis demonstrated an increase in non–breast cancer causes of death in patients that received RT.21 The Stockholm trial demonstrated an increase in cardiovascular deaths in patients that received photons to the left chest.31 A previous SEER registry–based study demonstrated an increase in cardiovascular deaths in women younger than 60 years of age who received adjuvant RT for left-sided breast cancer compared with patients treated for right-sided lesions.32 The diminished BCSS and OS observed in this study for medial tumor location highlights the importance of prospective randomized trials that target the IMN with RT. This study demonstrates that tumor site is an important prognostic factor in breast cancer. With the increasing ability to identify nodal drainage patterns to the IMNs, by lymphoscintigraphy and positron emission tomography scanning, and with the ability of conformal techniques to treat IMNs safely, these data have important implications for oncologists.
Supported in part by a grant from the Susan G. Komen Breast Cancer Foundation, Dallas, TX, and by contract no. NO1-PC-6700 from the National Cancer Institute, Bethesda, MD.
1. Zucali R, Mariani L, Marubini E, et al: Early breast cancer: Evaluation of the prognostic role of the site of the primary tumor. J Clin Oncol 16:1363–1366, 1998
2. Lohrisch C, Jackson J, Jones A, et al: Relationship between tumor location and relapse in 6,781 women with early invasive breast cancer. J Clin Oncol 18:2828–2835, 2000 3. Hammer J, Track C, Seewald DH: Breast cancer: The medial tumor location—An unfavorable disease! Results from 644 patients (1984–1995). Radiother Oncol 58:38, 2001 (suppl 1) 4. van der Ent FW, Kengen RA, van der Pol HA, et al: Halsted revisited: Internal mammary sentinel lymph node biopsy in breast cancer. Ann Surg 234:79–84, 2001[CrossRef][Medline] 5. Krag DN, Harlow S, Weaver D, et al: Radiolabelled sentinel node biopsy: Collaborative trial with the National Cancer Institute. World J Surg 25:823–828, 2001[CrossRef][Medline] 6. Uren RF, Howman-Giles R, Renwick SB, et al: Lymphatic mapping of the breast: Locating the sentinel lymph node. World J Surg 25:789–793, 2001[CrossRef][Medline] 7. Handley R: Natural history of breast cancer, in Harris J, Lippman ME, Morrow M, et al (eds): Diseases of the Breast. Philadelphia, PA, WB Saunders, 1996, p 364 8. Urban J: Is there a rationale for an extended radical procedure? Int J Radiat Oncol Biol Phys 2:985–988, 1977[Medline] 9. Lacour J, Le M, Caceres E, et al: Radical mastectomy versus radical mastectomy plus internal mammary dissection. Cancer 51:1941–1943, 1983[CrossRef][Medline] 10. Veronesi U, Valagussa P: Inefficacy of internal mammary node dissection in breast cancer surgery. Cancer 47:170–175, 1981[CrossRef][Medline]
11. Morrow M, Foster RS: Staging of breast cancer: A new rationale for internal mammary node biopsy. Arch Surg 116:748–751, 1981 12. Kaplan EL, Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 53:456–481, 1958 13. Host H, Brenvold IO, Loeb M: Post-operative radiotherapy in operable breast cancer: Long term results from the Oslo study. Int J Radiat Oncol Biol Phys 12:727–732, 1986[Medline] 14. Wallgren A, Arner O, Bergstrom J, et al: Radiation therapy in operable breast cancer: Results from the Stockholm trial on adjuvant therapy. Int J Radiat Oncol Biol Phys 12:533–537, 1986[Medline] 15. Arriagada R, Le MG, Mouriesse H, et al: Long-term effective internal mammary chain treatment: Results of a multivaried analysis of 1,195 patients with operable breast cancer and positive axillary nodes. Radiother Oncol 11:213–222, 1988[CrossRef][Medline]
16. Overgaard M, Hansen PS, Overgaard J, et al: Postoperative radiotherapy in high-risk premenopausal women with breast cancer who receive adjuvant chemotherapy. N Engl J Med 337:949–955, 1997 17. Overgaard M, Jensen M-B, Overgaard J, et al: Postoperative radiotherapy in high-risk postmenopausal women with breast cancer given adjuvant tamoxifen: Danish Breast Cancer Cooperative Group DBCG 82c randomised trial. Lancet 353:1641–1648, 1999[CrossRef][Medline]
18. Ragaz J, Stewart MJ, Le N, et al: Adjuvant radiotherapy and chemotherapy in node-positive premenopausal women with breast cancer. N Engl J Med 337:956–962, 1997 19. Early Breast Cancer Trialists’ Collaborative Group: Polychemotherapy for early breast cancer. An overview of the randomised trials. Lancet 352:930–942, 1998[CrossRef][Medline] 20. Early Breast Cancer Trialists’ Collaborative Group: Tamoxifen for early breast cancer. An overview of the randomised trials. Lancet 351:1451–1467, 1998[CrossRef][Medline] 21. Early Breast Cancer Trialists’ Collaborative Group: Favourable and unfavourable effects on long-term survival of radiotherapy for early breast cancer. An overview of the randomised trials. Lancet 355:1757–1770, 2000[CrossRef][Medline] 22. Recht A, Pierce SM, Abner A, et al: Regional nodal failure after conservative surgery and radiotherapy for early stage breast cancer. J Clin Oncol 9:988–996, 1991[Abstract] 23. Proulx GM, Lee RJ, Stomper PC: Internal mammary lymph node inclusion in standard tangent breast fields: Effects of body habitus. Breast J 7:111–116, 2001[CrossRef][Medline]
24. Bickell NA, Chassin MR: Determining the quality of breast cancer care: Do tumor registries measure up? Ann Intern Med 132:705–710, 2000
25. Malin JL, Kahn KL, Adams J, et al: Validity of cancer registry data for measuring the quality of breast cancer care. J Natl Cancer Inst 94:835–844, 2002
26. Percy C, Stanek E, Gloeckler L: Accuracy of cancer death certificates and its effect on cancer mortality statistics. Am J Public Health 71:242–250, 1981 27. Du X, Freeman JL, Goodwin JS: Information on radiation treatment in patients with breast cancer: The advantages of the linked Medicare and SEER data. J Clin Epidemiol 52:463–470, 1999[CrossRef][Medline] 28. Freedman GM, Fowble BL, Nicolaou N, et al: Should internal mammary lymph nodes in breast cancer be a target for the radiation oncologist? Int J Radiat Oncol Biol Phys 46:805–814, 2000[CrossRef][Medline] 29. Hojris I, Overgaard M, Christensen JJ, et al: Morbidity and mortality of ischaemic heart disease in high-risk breast cancer patients after adjuvant postmastectomy systemic treatment with or without radiotherapy: Analysis of DBCG 82b and 82c randomised trials. Radiotherapy Committee of the Danish Breast Cancer Cooperative Group. Lancet 354:1425–1430, 1999[CrossRef][Medline]
30. Nixon AJ, Manola J, Gelman R, et al: No long-term increase in cardiac-related mortality after breast-conserving surgery and radiation therapy using modern techniques. J Clin Oncol 16:1374–1379, 1998 31. Rutqvist LE, Lax I, Fornander T, et al: Cardiovascular mortality in a randomised trial of adjuvant radiation therapy versus surgery alone in primary breast cancer. Int J Radiat Oncol Biol Phys 22:887–896, 1992[Medline] 32. Paszat LF, Mackillop WJ, Groome PA, et al: Mortality from myocardial infarction after adjuvant radiotherapy for breast cancer in the Surveillance, Epidemiology, and End-Results cancer registries. J Clin Oncol 16:2625–2631, 1998[Abstract] Submitted December 11, 2001; accepted October 16, 2002.
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2003 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
