|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 25, No 31 (November 1), 2007: pp. 4952-4960 © 2007 American Society of Clinical Oncology. DOI: 10.1200/JCO.2006.08.0499 Overall Survival and Cause-Specific Mortality of Patients With Stage T1a,bN0M0 Breast Carcinoma
From the Departments of Breast Medical Oncology and Biostatistics, The University of Texas M.D. Anderson Cancer Center, Houston, TX; and the Department of Surgical Oncology, Institut Gustave Roussy, Villejuif, France Address reprint requests to Emer O. Hanrahan, MD, The University of Texas M.D. Anderson Cancer Center, Unit 10, 1515 Holcombe Blvd, Houston, TX 77030; e-mail: ehanraha{at}mdanderson.org
Purpose With mammographic screening, the frequency of diagnosis of stage T1a,bN0M0 breast cancer has increased. Prognosis after locoregional therapy and benefit from adjuvant systemic therapy are poorly defined. We reviewed T1a,bN0M0 breast cancer cases registered in the Surveillance, Epidemiology, and End Results (SEER) Program to investigate the impact of prognostic factors on breast cancer–specific (BCSM) and non–breast cancer–related mortality. Methods We identified T1a,bN0M0 breast cancer cases registered in the SEER Program from 1988 to 2001, and used the Kaplan-Meier product limit method to describe overall survival (OS). We estimated the probabilities of death resulting from breast cancer and from other causes, and analyzed associations of patient and tumor characteristics with OS, BCSM, and non–breast cancer–related mortality using the log-rank test, Cox proportional hazards models, and a competing-risk model. We constructed nomograms to assist physicians in adjuvant therapy decision making. Results We identified 51,246 T1a,bN0M0 cases. Median follow-up was 64 months (range, 1 to 167 months). Median age at diagnosis was 65 years (range, 20 to 101 years). Ten-year probabilities of all-cause mortality and BCSM were 24% and 4%, respectively. Characteristics associated with increased probability of BCSM included age younger than 50 years at diagnosis, high tumor grade, estrogen receptor–negative status, progesterone receptor–negative status, and fewer than six nodes removed at axillary dissection. The constructed nomograms allow a comparison of predicted breast cancer–specific survival and non-breast cancer–specific survival in individual patients. Conclusion Overall, the prognosis of patients with T1a,bN0M0 breast cancer is excellent. However, subgroups of patients who are at higher risk of BCSM and who should be considered for adjuvant systemic therapy can be identified.
A stage T1a,bN0M0 breast cancer is a stage I breast cancer with a maximum tumor diameter of 1 cm, where T1a is no more than 0.5 cm and T1b is 0.6 to 1.0 cm.1 The Surveillance, Epidemiology, and End Results (SEER) Program is a network of 14 population-based and three supplemental cancer registries covering approximately 26% of the US population. Between 1990 and 1998, the rate of T1 tumors diagnosed among women aged 50 to 69 years in nine SEER Program registries increased from 143.5/100,000 to 163.5/100,000.2 Almost two thirds of 171,479 breast cancer cases in the SEER Program between 1996 and 2000 were node negative.3 The increase in stage I breast cancers has been attributed to mammographic screening.4-6 However, the clinical course of T1a,bN0M0 cancer after locoregional therapy alone remains poorly defined.7 The extent of benefit for patients from adjuvant systemic therapy (ADST) depends largely on their baseline risk of recurrence and risk of dying as a result of other causes. A number of published retrospective reviews, mostly from single institutions, describe heterogeneous relapse-free survival rates for T1a,bN0M0 breast cancer patients, ranging from 87% at 5 years to more than 90% at 10 years.8-14 The SEER database does not provide information on recurrence, but it records cause of death. Examining outcome in such a large population-based cohort yields survival results that are likely more accurate than single-institution reports. The only report using SEER data to consider competing causes of death among patients with breast cancer did not specifically examine T1a,bN0M0 cases.15 To better define the prognosis of patients with T1a,bN0M0 breast cancer and to assist physicians with ADST decisions, we reviewed all T1a,bN0M0 cases registered in the SEER Program from 1988 to 2001 and analyzed their overall survival (OS), breast cancer–specific mortality (BCSM), and probability of death resulting from other causes (considered a competing risk). We also evaluated the associations of patient and disease characteristics with outcome and built nomograms examining breast cancer–specific survival (BCSS) and non–breast cancer–specific survival (non-BCSS; survival from all potential causes of death excluding breast cancer).
Using SEER registry public use data tapes from 12 registries, we identified patients diagnosed with T1a,bN0M0 breast cancer from January 1, 1988, to December 31, 2001, and tabulated their characteristics. We excluded patients who were registered but for whom no follow-up data were recorded (Appendix, online only) and patients who had sentinel lymph node biopsy (SLNB). SLNB cases were described only after 1998, and therefore did not have prolonged follow-up. Fifty-six histology codes were represented and collapsed into nine categories by a pathologist (adenocarcinoma not otherwise specified, invasive ductal carcinoma [IDC], infiltrating lobular carcinoma, medullary, metaplastic, mixed, mucinous, Paget's, and other). SEER surgery codes were changed after 1998.16,17 We reviewed and collapsed surgery codes for the primary site into four categories: no surgery, breast-conserving therapy (BCT), mastectomy, and unknown. When we considered whether patients had axillary lymph node (ALN) surgery, we noted inconsistencies between surgical procedure coding and the number of examined ALNs. Therefore, patients recorded as having zero ALNs examined were considered not to have had ALN surgery, and patients with more than zero ALNs examined were considered to have had ALN surgery. Extent of axillary surgery was categorized as either zero, one to five, or six or more ALNs examined at ALN dissection (ALND). We chose six ALNs as a cutoff based on American Joint Committee on Cancer (AJCC) staging recommendations.1 The SEER Program classifies tumor grade as I (well differentiated), II (moderately differentiated), III (poorly differentiated), and IV (undifferentiated). Median follow-up was calculated as the median observed survival time among all patients. OS was measured as the time from diagnosis to death, date of last follow-up, or December 21, 2001, if date of last follow-up was after December 21, 2001. We used the Kaplan-Meier product limit method to describe OS and the log-rank test to assess for differences between patient groups.18,19 We fitted Cox models to assess the multivariable relationships between patient characteristics and OS.20 Estrogen receptor (ER) and progesterone receptor (PR) status were recorded only for patients diagnosed after 1990 and were associated with a large number of missing values. Therefore, two Cox proportional hazards models were fitted: one including and one excluding ER/PR. For each model, we tested for pairwise interactions specified a priori; all possible interactions between grade, tumor size, histology, and ER (if present). We used the likelihood ratio test to determine the statistical significance of interaction terms. The proportional hazards assumption was assessed graphically with residual plots and tested according to the method of Grambsch and Therneau.21
We classified cause of death as either breast cancer or another cause. In the analysis of BCSM, deaths from other causes were considered a competing risk. We used the methods described in Gooley et al22 and Pepe and Mori23 to estimate probability of death resulting from breast cancer or other causes, and to test for associations between patient characteristics and death resulting from breast cancer. We present the probability of death resulting from breast cancer along with the probability of death resulting from other causes. Note that if the probability of death from another cause is high, then the probability of death from breast cancer must be low because the sum of these two quantities cannot exceed the overall probability of death. The test statistics are based on cumulative weighted differences between the two groups. Differences at earlier time points are given more weight than differences at later time points that have fewer patients because of exclusion. The method of Pepe and Mori only allows for comparisons between two groups and, therefore, race, grade, histology, ALN surgery, and number of ALNs examined (< 6 or A Cox model with the dichotomized prognostic factors was used to construct nomograms to predict individual patient probabilities of BCSS and non-BCSS at 5 and 10 years. To determine the cause-specific regression analysis through Cox's model, patients with events not related to breast cancer were excluded at their event times. The models were built with age, ER and PR status, grade, T size, and axillary surgery as variables, and the final models include only those variables found to be significantly associated with outcome in these analyses. Age was fitted using restricted cubic splines to relax the linearity assumptions. The nomograms were internally validated by bootstrapping with 500 resamples as a means of calculating a relatively unbiased measure of ability to discriminate between patients as quantified by the concordance index (similar to the area under the receiver operating curve).24 Subsequently, the predicted probabilities of BCSS and non-BCSS were compared with the actual outcomes using 500 bootstrap resamples to decrease overfit bias.
Overall Survival A total of 51,246 cases of T1a,bN0M0 breast cancer were eligible for inclusion in this analysis (Appendix, online only), with 39.8% of these cases registered between 1997 and 2001. Table 1 summarizes available patient and tumor characteristics. Median patient age was 65 years. Median tumor size was 8 mm.
Median follow-up for these patients was 64 months (range, 1 to 167 months). The proportion remaining alive at 5 and 10 years was 0.902 (95% CI, 0.899 to 0.905) and 0.759 (95% CI, 0.753 to 0.765), respectively (Fig A1, online only). Factors associated with prolonged OS included age younger than 50 years at diagnosis, non–African American race, T1a rather than T1b tumors, and positive PR status (Table 1). Patients who had an ALND had longer OS than those who had no ALN surgery, and OS was longer with greater number of ALNs examined. Table 2 shows results of the two Cox proportional hazards models. Because of missing values, models 1 and 2 include 37,047 and 26,873 observations, respectively. Variables that were significantly associated with OS in the univariate analysis tended to remain significantly associated with OS in the multivariable model. In model 1, a 1-year increase in age at diagnosis was associated with 1.06 times the risk of death. A 1-mm increase in tumor size was associated with 1.02 times the risk of death. Compared with no axillary surgery, ALND with one to five or six or more nodes examined was associated with a reduction in the risk of death by approximately one quarter or one half, respectively. Inference was similar for model 2. In addition, ER-positive disease but not PR-positive disease was associated with significantly decreased risk of death (hazard ratio = 0.76). No interactions were found to be statistically significant in either model.
BCSM and Competing-Risk Analysis Among the 51,246 cases, there were 1,340 deaths resulting from breast cancer and 5,931 from other causes. Overall, probability of death from breast cancer was significantly lower than that from other causes and increased at a much slower rate (Fig A2, online only). At 5 and 10 years after diagnosis, probability of death resulting from breast cancer was 0.02 and 0.04, respectively, and probability of death from other causes was 0.08 and 0.20, respectively. The 10 most common causes of death in descending order of frequency were heart disease (n = 1,727), breast cancer, other (n = 633), cerebrovascular diseases (n = 519), cancer of lung/bronchus (n = 375), no cause listed (n = 363), chronic obstructive pulmonary disease (n = 306), pneumonia/influenza, miscellaneous cancer, and diabetes mellitus. Estimates of probabilities of death from breast cancer and other causes by patient and tumor characteristics are presented in Table 3. Patients who were less than 50 years of age at diagnosis had a greater probability of death resulting from breast cancer (P = .0005) and a significantly lower probability of death resulting from other causes compared with patients who were 50 years of age or older (P < .0001; Fig 1A). Probability of death resulting from breast cancer was similar between patients with a tumor size of 5 mm or less and patients with tumors 6 mm or larger (Fig 1B). Although patients had similar OS by ER status, patients with ER-negative disease were more likely to die as a result of breast cancer than were those with ER-positive disease, and they were less likely to die as a result of other causes than were those with ER-positive disease (P < .0001 for both; Fig 1C). Patients with PR-negative disease were also more likely to die as a result of breast cancer. Those with grade 1 to 2 disease were less likely to die as a result of breast cancer compared with patients with grade 3 to 4 disease, and they were more likely to die as a result of other causes than were patients with grade 3 to 4 disease, consistent with the similar OS of the two groups (Fig 1D). Patients who received radiation, ALN surgery, or had at least six ALNs examined were less likely to die as a result of breast cancer than were patients who received no radiation, no lymph node surgery, or had fewer than six ALNs examined, respectively. Patients who received BCT, radiation, ALN surgery, or had at least 6 ALNs examined were also less likely to die as a result of non–breast cancer causes.
We stratified by age at diagnosis and analyzed probabilities of death resulting from breast cancer and other causes by patient and tumor characteristics (Table A1, online only). There were 8,567 patients aged less than 50 years at diagnosis. Among all deaths in this group, there was a statistical trend to more deaths resulting from breast cancer (n = 270) than from other causes (n = 142), and this held across all patient and tumor characteristics. Among 42,679 patients aged at least 50 years at diagnosis, 1,070 patients died from breast cancer and 5,784 patients died from other causes (P < .0001). In contrast to younger patients, probability of death from breast cancer was significantly lower than that from other causes across all characteristics. Factors that were significantly related to probability of death from breast cancer among both patient groups were tumor grade 3 to 4, ER-/PR-negative disease, and fewer than six ALNs examined (P  .0001). Those aged less than 50 years with grade 3 to 4 tumors had a 10-year probability of death resulting from breast cancer of 11%, compared with a 3% probability of death resulting from other causes. For patients at least 50 years of age, the corresponding probabilities of death from breast cancer or from other causes were 6% and 22%, respectively. For those aged less than 50 years with ER-negative tumors, the 10-year probability of breast cancer death was 10% and the probability of death resulting from other causes was 3%, whereas the corresponding probabilities for older patients were 7% and 18%, respectively (Fig 2).
Nomograms Because of the limits of the previous statistical analysis in terms of the number of factors that could be analyzed at one time, we constructed nomograms to predict individual patient probability of BCSS and non-BCSS at 5 and 10 years on the basis of multiple patient- and tumor-specific factors (Fig 3A and 3B). These nomograms allow for easy, simultaneous consideration of prognostic factors and competing risks. They can help physicians identify patients with T1a,bN0M0 breast cancer who may have a poor outcome in terms of BCSS, and who would likely benefit from ADST.
We have shown that patients with T1a,bN0M0 breast cancers registered in the SEER Program from 1988 to 2001 had an excellent BCSS. Their 10-year overall mortality rate was 24%, but their 10-year BCSM was only 4%. Thus, these patients had a five-fold higher risk of dying from causes other than breast cancer. Most published retrospective reviews of outcome in T1a,bN0M0 breast cancer are single-institution experiences and results are heterogeneous.8-14,25-36 Outcome reports from population-based cohorts have advantages of much larger patient numbers and results that are likely to be more generally applicable. Although the SEER Program does not provide data about disease recurrence, the recorded cause of death can allow crude cumulative cause-specific probabilities of death based on a competing-risk model to calculate measures of cause-specific mortality. We have used this method of analysis in the current assessment of the SEER data. This provides a better measure of prognosis after breast cancer diagnosis than overall or relative survival, and may be especially helpful when considering the risk:benefit ratio of ADST. We also constructed nomograms allowing comparison for individual patients of predicted BCSS and non-BCSS. Nomograms are extremely useful tools to allow individualization of ADST decisions.37 To our knowledge, such analyses have not previously been published for T1a,bN0M0 breast cancer. Although the Adjuvant! Online program (http://www.adjuvantonline.com) can be used to determine the risk of BCSS in T1a,bN0M0 cases, it does not make any distinction based on tumor size within this subgroup.37 Schairer et al15 were the first group to perform a detailed assessment of breast cancer outcome using a competing-risk model. They analyzed the vital status and cause of death for 430,510 patients with localized, node-negative breast cancer registered from 1973 to 2000 in nine SEER registries and found that probability of death resulting from breast cancer exceeded that resulting from all other causes in patients aged less than 50 years. We similarly found a trend toward higher probability of death resulting from breast cancer than from other causes for patients aged less than 50 years at diagnosis of T1a,bN0M0 breast cancer, and we identified subgroups with adverse features (ER- or PR-negative disease, tumor grade 3 to 4) that are associated with a particularly increased risk of BCSM. Schairer et al15 also reported increased BCSM for cases with ER-negative disease, but they did not consider tumor grade or PR status in their analysis. Although Schairer et al found that probability of death from breast cancer for patients diagnosed with localized or regional disease (stage I-III by AJCC system1) was significantly greater in African American compared with white patients, this was not the case in our study confined to stage T1a,bN0M0 disease. In our analysis, characteristics of T1a,bN0M0 disease associated with elevated probability of BCSM included younger age at diagnosis, high tumor grade, and ER- and PR-negative status. This is consistent with the results of multiple retrospective reviews.8,9,25,28-31,33,38 Our finding that a 1-mm increase in tumor size was associated with 1.02 to 1.03 times the risk of death is consistent with another SEER data analysis that showed increasing crude cumulative death rate from 10% to 25% with increasing tumor size from 3 to 50 mm in node-negative cases.39 Inadequate ALN assessment (< 6 nodes examined) resulted in higher probability of death from breast cancer, which is likely caused by understaging in a proportion of these patients. The prognostic importance of number of nodes resected at ALND has similarly been shown in institution-based reviews and in a recent SEER analysis.40-42 We further found that patients who did not have breast-conserving therapy, radiation therapy, or inadequate ALN assessment were at a significantly greater risk of death from other causes. This may reflect greater comorbidities in some of these patients that compromised their ability to tolerate optimal locoregional management and increased their risk of non–breast cancer–related death. There are limitations to our study. Most importantly, information on ADST use is not available in the SEER database. The proportion of patients who received adjuvant chemotherapy or hormonal therapy is unknown, and the BCSS calculated from SEER data may somewhat overestimate the outcome that would be found in an entirely untreated population. Data on tumor recurrence is not provided in the SEER database. However, evidence suggests that the accuracy of cause-of-death coding for cancer patients on death certificates and in SEER is high.43-46 Although the range of follow-up is long, the median is only 64 months. Nevertheless, because of the large number of patients available from SEER and the fact that 76% of patients are still at risk at 10 years, we can estimate 10-year survival with a high degree of confidence. Hormone receptor status was not measured in a central laboratory, and is not standardized in the population. Information on some other prognostic factors, such as lymphovascular invasion and human epidermal growth factor 2 (HER-2) status, is unavailable. Our nomograms allow for calculation of the probability of BCSS based on individual characteristics, but they do not consider the extent of comorbidities in the calculation of probability of non-BCSS. The Adjuvant! Online program can be used in conjunction with our nomograms to incorporate comorbidities into the estimation of individual non-BCSS probability. Our data are derived from a large, population-based cohort and have a number of practical or clinical implications. T1a,bN0M0 cases overall have an excellent prognosis, but the nomograms will help clinicians identify who may be at higher risk of BCSM and assist in ADST decision making. Because most of these patients die as a result of other causes, many of which are at least partly lifestyle related (eg, cardiovascular), a healthy lifestyle and the management of non–breast cancer–related health problems in these patients are important.
The author(s) indicated no potential conflicts of interest.
Conception and design: Emer O. Hanrahan, Ana M. Gonzalez-Angulo, Gabriel N. Hortobagyi, Vicente Valero Collection and assembly of data: Emer O. Hanrahan, Ana M. Gonzalez-Angulo, Sharon H. Giordano Data analysis and interpretation: Emer O. Hanrahan, Ana M. Gonzalez-Angulo, Sharon H. Giordano, Roman Rouzier, Kristine R. Broglio Manuscript writing: Emer O. Hanrahan, Ana M. Gonzalez-Angulo, Kristine R. Broglio, Roman Rouzier, Gabriel N. Hortobagyi, Vicente Valero Final approval of manuscript: Emer O. Hanrahan, Ana M. Gonzalez-Angulo, Sharon H. Giordano, Roman Rouzier, Kristine R. Broglio, Gabriel N. Hortobagyi, Vicente Valero
Patients A total of 57,835 cases of T1a,bN0M0 breast cancer were identified. Patients with errors in OS calculation (n = 5), with follow-up of 0 months (n = 471), or with SLNB (n = 6,317) were excluded. Cases with follow-up of 0 months represent patients whose diagnosis was recorded but who either were then lost to follow-up or were reported to the database within 1 month of the cutoff date for this analysis (453 of 471, 96.2%), or who died within a month of diagnosis (18 of 471, 3.8%). There were 204 cases with both SLNB and a follow-up of 0 months.
Nomograms
Supported by the Susan G. Komen Breast Cancer Foundation and the Nellie B. Connally Breast Cancer Research Fund. Presented in part at the 41st Annual Meeting of the American Society of Clinical Oncology, May 13-17, 2005, Orlando, FL. Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
1. Greene FL, Page DL, Fleming ID, et al: AJCC Cancer Staging Handbook: TNM Classification of Malignant Tumors (ed 6). New York, NY, John Wiley & Sons, 2002 2. Schootman M, Jeffe D, Reschke A, et al: The full potential of breast cancer screening use to reduce mortality has not yet been realized in the United States. Breast Cancer Res Treat 85 : 219 -222, 2004[CrossRef][Medline] 3. Jemal A, Clegg LX, Ward E, et al: Annual report to the nation on the status of cancer, 1975-2001, with a special feature regarding survival. Cancer 101 : 3 -27, 2004[CrossRef][Medline] 4. Luke C, Nguyen AM, Priest K, et al: Female breast cancers are getting smaller, but socio-demographic differences remain. Aust N Z J Public Health 28 : 312 -316, 2004[Medline] 5. Fracheboud J, Otto SJ, van Dijck JA, et al: National Evaluation Team for Breast Cancer Screening (NETB): Decreased rates of advanced breast cancer due to mammography screening in the Netherlands. Br J Cancer 91 : 861 -867, 2004[Medline] 6. Vacek PM, Geller BM, Weaver DL, et al: Increased mammography use and its impact on earlier breast cancer detection in Vermont, 1975-1999. Cancer 94 : 2160 -2168, 2002[CrossRef][Medline] 7. Hanrahan EO, Valero V, Gonzalez-Angulo AM, et al: Prognosis and management of patients with node-negative invasive breast carcinoma that is 1 cm or smaller in size (stage 1; T1a,bN0M0): A review of the literature. J Clin Oncol 24
: 2113
-2122, 2006 8. Rosen PP, Groshen S, Saigo PE, et al: A long-term follow-up study of survival in stage I (T1N0M0) and stage II (T1N1M0) breast carcinoma. J Clin Oncol 7 : 355 -366, 1989[Abstract] 9. Rosen PP, Groshen S, Kinne DW, et al: Factors influencing prognosis in node-negative breast carcinoma: Analysis of 767 T1N0M0/T2N0M0 patients with long-term follow-up. J Clin Oncol 11
: 2090
-2100, 1993 10. Moon TE, Jones SE, Bonadonna G, et al: Development and use of a natural history database of breast cancer studies. Am J Clin Oncol 10 : 396 -403, 1987[Medline] 11. Tinnemans J, Wobbes T, Holland R, et al: Treatment and survival of females with non-palpable breast carcinoma. Ann Surg 209 : 249 -253, 1989[Medline] 12. Stierer M, Rosen HR, Weber R, et al: Long term analysis of factors influencing the outcome in carcinoma of the breast smaller than one centimeter. Surg Gynecol Obstet 175 : 151 -160, 1992[Medline] 13. Leitner SP, Swern AS, Weinberger D, et al: Predictors of recurrence for patients with small (one centimeter or less) localized breast cancer (T1a,b N0 M0). Cancer 76 : 2266 -2274, 1995[CrossRef][Medline] 14. Fisher B, Dignam J, Tan-Chiu E, et al: Prognosis and treatment of patients with breast tumors of one centimeter or less and negative axillary lymph nodes. J Natl Cancer Inst 93
: 112
-120, 2001 15. Schairer C, Mink PJ, Carroll L, et al: Probabilities of death from breast cancer and other causes among female breast cancer patients. J Natl Cancer Inst 96
: 1311
-1321, 2004 16. Shambaugh EM, Weiss MA, Axtell LM, et al: Summary staging guide for the cancer Surveillance, Epidemiology, and End Results reporting program. Bethesda, MD, National Institutes of Health, 1983 17. Young JL Jr, Roffers SD, Ries LAG, et al (eds): SEER summary staging manual—2000: Codes and coding instructions. NIH Pub. No. 01-4969. http://seer.cancer.gov/tools/ssm/breast_femgen.pdf 18. Kaplan EL, Meier P: Non-parametric estimation from incomplete observations. J Am Stat Assoc 53 : 457 -481, 562-563, 1958[CrossRef] 19. Harrington DP, Fleming TR: A class of rank test procedures for censored survival data. Biometrika 69
: 553
-566, 1982 20. Cox DR: Regression models and life tables (with discussion). J R Stat Soc B 34 : 187 -220, 1972 21. Grambsch P, Therneau T: Proportional hazards tests and diagnostics based on weighted residuals. Biometrika 81
: 515
-526, 1994 22. Gooley TA, Leisenring W, Crowley J, et al: Estimation of failure probabilities in the presence of competing risks: New representations of old estimators. Stat Med 18 : 695 -706, 1999[CrossRef][Medline] 23. Pepe MS, Mori M: Kaplan-Meier: Marginal or conditional probability curves in summarizing competing risks failure time data? Stat Med 112 : 737 -751, 1993 24. Harrell FE Jr, Lee KL, Mark DB: Multivariable prognostic models: Issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 15 : 361 -387, 1996[CrossRef][Medline] 25. Stierer M, Rosen HR, Weber R, et al: Long term analysis of factors influencing the outcome in carcinoma of the breast smaller than one centimeter. Surg Gynecol Obstet 175 : 151 -160, 1992[Medline] 26. Ichizawa N, Fukutomi T, Iwamoto E, et al: Long-term results of T1a, T1b and T1c invasive breast carcinomas in Japanese women: Validation of the UICC T1 subgroup classification. Jpn J Clin Oncol 32
: 108
-109, 2002 27. Quiet CA, Ferguson DJ, Weichselbaum RR, et al: Natural history of node-negative breast cancer: A study of 826 patients with long-term follow-up. J Clin Oncol 13 : 1144 -1151, 1995[Abstract] 28. Colleoni M, Rotmensz N, Peruzzotti G, et al: Minimal and small size invasive breast cancer with no axillary lymph node involvement: The need for tailored adjuvant therapies. Ann Oncol 15
: 1633
-1639, 2004 29. Lee AK, Loda M, Mackarem G, et al: Lymph node negative invasive breast carcinoma 1 centimeter or less in size (T1a,bN0M0): Clinicopathologic features and outcome. Cancer 79 : 761 -771, 1997[CrossRef][Medline] 30. Rosner D, Lane WW: Should all patients with node-negative breast cancer receive adjuvant therapy? Identifying additional subsets of low-risk patients who are highly curable by surgery alone. Cancer 68 : 1482 -1494, 1991[CrossRef][Medline] 31. Rosner D, Lane WW: Predicting recurrence in axillary-node negative breast cancer patients. Breast Cancer Res Treat 25 : 127 -139, 1993[CrossRef][Medline] 32. Crowe JP Jr, Gordon NH, Shenk RR, et al: Primary tumor size: Relevance to breast cancer survival. Arch Surg 127
: 910
-915, 1992 33. Joensuu H, Pylkkanen L, Toikkanen S: Late mortality from pT1N0M0 breast carcinoma. Cancer 85 : 2183 -2189, 1999[CrossRef][Medline] 34. Arnesson LG, Smeds S, Fagerberg G: Recurrence-free survival in patients with small breast cancer: An analysis of cancers 10 mm or less detected clinically and by screening. Eur J Surg 160 : 271 -276, 1994[Medline] 35. Wood WC, Anderson M, Lyles RH, et al: Can we select which patients with small breast cancers should receive adjuvant chemotherapy? Ann Surg 235 : 859 -862, 2002[CrossRef][Medline] 36. Mann GB, Port ER, Rizza C, et al: Six-year follow-up of patients with microinvasive, T1a, and T1b breast carcinoma. Ann Surg Oncol 6 : 591 -598, 1999[CrossRef][Medline] 37. Ravdin PM, Siminoff LA, Davis GJ, et al: Computer program to assist in making decisions about adjuvant therapy for women with early breast cancer. J Clin Oncol 19
: 980
-991, 2001 38. Chia SK, Speers CH, Bryce CJ, et al: Ten-year outcomes in a population-based cohort of node-negative, lymphatic, and vascular invasion-negative early breast cancers without adjuvant systemic therapies. J Clin Oncol 22
: 1630
-1637, 2004 39. Verschraegen C, Vinh-Hung V, Cserni G, et al: Modeling the effect of tumor size in early breast cancer. Ann Surg 241 : 309 -318, 2005[CrossRef][Medline] 40. Joslyn SA, Konety BR: Effect of axillary lymphadenectomy on breast carcinoma survival. Breast Cancer Res Treat 91 : 11 -18, 2005[CrossRef][Medline] 41. Petrik DW, McCready DR, Sawka CA, et al: Association between extent of axillary lymph node dissection and patient, tumor, surgeon, and hospital factors in patients with early breast cancer. J Surg Oncol 82 : 84 -90, 2003[CrossRef][Medline] 42. Blancas I, Garcia-Puche JL, Bermejo B, et al: Low number of examined lymph nodes in node-negative breast cancer patients is an adverse prognostic factor. Ann Oncol 17
: 1644
-1649, 2006 43. Kircher T, Nelson J, Burdo H: The autopsy as a measure of accuracy of the death certificate. N Engl J Med 313 : 1263 -1269, 1985[Abstract] 44. Percy C, Stanek E III, Gloeckler L: Accuracy of cancer death certificates and its effect on cancer mortality statistics. Am J Public Health 71
: 242
-250, 1981 45. Percy CL, Miller BA, Gloeckler Ries LA: Effect of changes in cancer classification and the certificates on trends in cancer mortality. Ann N Y Acad Sci 609 : 87 -99, 1990[Medline] 46. Brinkley D, Haybittle JL, Alderson MR: Death certification in cancer of the breast. BMJ 289
: 465
-467, 1984 Submitted July 26, 2006; accepted August 7, 2007.
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2007 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
6) were collapsed. All P values are two sided. On the basis of the large number of univariate tests, only P values less than .0002 were considered statistically significant. 
