Life Expectancy of Screen-Detected Invasive Breast Cancer Patients Compared With Women Invited to the Nijmegen Screening Program

Transcription

1 Original Article Life Expectancy of Screen-Detected Invasive Breast Cancer Patients Compared With Women Invited to the Nijmegen Screening Program Johannes D. M. Otten 1,2 ; Mireille J. M. Broeders 1,2,3 ; Gerard J. Den Heeten, MD 2,3 ; Roland Holland, MD 3 ; Jacques Fracheboud, MD 3,4 ; Harry J. De Koning, MD 2,4 ; and André L. M. Verbeek, MD 1,2 BACKGROUND: Screening can lead to earlier detection of breast cancer and thus to an improvement in survival. The authors studied the life expectancy of women with screen-detected invasive breast cancer (patients) compared with women invited to the breast cancer screening program in Nijmegen, the Netherlands (comparison group). METHODS: Each patient diagnosed between 1975 and 2006 was randomly age-matched with a woman invited in the same calendar year and free from breast cancer at the time of diagnosis of the patient. Survival analyses were performed to study differences in life expectancy. RESULTS: The life expectancy for 858 patients was 6 years shorter than for the comparison group. However, for 360 patients with small (<15 mm) invasive breast cancer, life expectancy was similar to that of the comparison group. In contrast, for patients detected with larger tumors (15 mm) the life expectancy was 6 to 12 years shorter, depending on tumor size. Furthermore, life expectancy was modified by screening history. For patients who had a negative screening examination 2 years before the detection of their breast cancer, the difference in life expectancy from the comparison group became smaller for the larger tumor sizes (15 mm). CONCLU- SIONS: In conclusion, about 40% (360 of 858) of all women with invasive screen-detected breast cancer have the same life expectancy as women from the comparison group (reflecting the general population). For women diagnosed with larger tumors at diagnosis, life expectancy diminishes with increasing tumor size and is modified by screening history. Cancer 2010;116: VC 2009 American Cancer Society. KEYWORDS: mass screening, breast neoplasms, mammography, survival, life expectancy. Breast cancer is the most common cancer and a leading cause of death among women worldwide. 1 In many countries, mammographic breast cancer screening programs have been implemented to reduce breast cancer mortality. 2 In the Netherlands, breast cancer mortality showed a marked decline after implementation of the screening program. 3 Despite this favorable community perspective, individual patients with screen-detected breast cancer still face the threat of dying from this disease. At the time of breast cancer detection, the average number of life years left is still substantial, ranging from about 10 to 30 years depending on the age of the woman. Long-term survival studies usually describe differences between distinctive groups of patients or between different prognostic factors in 1 subgroup of patients. 4,5 To study longevity, we compare survival from screen-detected patients with survival from apparently healthy women invited to screening, representing the general population. We further take tumor size into account, because it influences survival and mortality 6 ; the longer the tumor has been growing, the more likely is the metastatic spread. 7 Furthermore, the impact of mammographic screening is based on detection of breast cancer with a smaller tumor size. This study was performed with individual data from 1 of the world s longest running screening programs in Nijmegen, the Netherlands. We analyzed the life expectancy of women with screen-detected invasive breast cancer stratified by tumor size (patients) compared with women invited to the breast cancer screening program (comparison group). Corresponding author: Johannes D. M. Otten, Department of Epidemiology, Biostatistics, and HTA (133), Radboud University Nijmegen Medical Center, P.O. Box 9101, 6500 HB Nijmegen, the Netherlands; Fax: (011) ; j.otten@ebh.umcn.nl 1 Department of Epidemiology, Biostatistics, and Health Technology Assessment, Radboud University Nijmegen Medical Centre, Nijmegen, the Netherlands; 2 National Evaluation Team for Breast Cancer Screening in the Netherlands, Erasmus Medical Center, University Medical Center Rotterdam, Rotterdam, the Netherlands; 3 National Expert and Training Center for Breast Cancer Screening, Nijmegen, the Netherlands; 4 Department of Public Health, Erasmus Medical Center, University Medical Center Rotterdam, Rotterdam, the Netherlands DOI: /cncr.24771, Received: March 16, 2009; Revised: May 19, 2009; Accepted: May 26, 2009, Published online December 7, 2009 in Wiley InterScience ( 586 Cancer February 1, 2010

2 Screen-Detected Breast Cancer Longevity/Otten et al MATERIALS AND METHODS Study population Before the introduction of the nationwide biennial breast cancer screening program in the Netherlands in 1989, a demonstration program had already been started in the city of Nijmegen in The results of the first 9 rounds of the Nijmegen program have been published in detail elsewhere. 8 In this study, covering the period 1975 to 2006, data were collected on the number of women living in Nijmegen, invited for screening, referred on the basis of a positive screening examination, and diagnosed with primary breast cancer. Between 1975 and 2006, 384,000 invitations were sent to 53,085 women aged 35 years and older. The interval between 2 invitations for screening was 2 years and did not change over time. About 241,500 screening examinations were performed, resulting in 2309 women referred for further diagnostic workup, of whom 1290 women were false-positively referred based on imaging and/or histology. Eight women refused further clinical workup and were excluded for lack of histological confirmation of breast cancer. For the remaining 1011 referred women, 153 women were detected with carcinoma in situ and 858 with primary invasive breast cancer. Because breast cancer in situ, correctly classified and treated, rarely ends in breast cancer death, 9 no difference in life expectancy between these patients and the comparison group is expected. For this reason, we only included invasive screen-detected breast cancer patients for further analysis. We defined the comparison group in this study as the apparently healthy women invited to the Nijmegen breast cancer screening program. Furthermore, because factors such as year of detection and age strongly influence survival, we randomly matched each patient to a comparison woman (1:1) by the patient s age and year of detection. At this particular calendar year, the comparison woman had to be free from breast cancer. If a comparison woman was diagnosed with breast cancer through screening later on, she remained in the comparison group for her previously matched patient. However, she was also added to the screen-detected cancer group and matched with a comparison woman for her calendar year of diagnosis. For 1 patient there was no proper match, leaving 857 comparison women. Informed consent was obtained for all women who participated in the Nijmegen breast cancer screening program. We received information from the Municipal Personal Records Data Base on vital status (death or migration) of women eligible for the Nijmegen breast cancer screening program on a monthly base from the start of the study in 1975 until This information was merged for the patients and for the pool of invited women eligible for the comparison group. In this way, no difference in information gathering across both groups was present. Women who were alive were withdrawn at December 31, 2007, and women who left Nijmegen were censored at date of migration. The start date for follow-up was the date of (invitation for) the screening examination at which the patient s cancer was detected or the matched date of screening invitation for the comparison woman. Follow-up time in years (average year length: days) was calculated from date of invitation up to date of death or migration, or to the end of the study. Data Analysis Kaplan-Meier survival curves for patients and comparison women were used to determine life expectancy (median survival time) in both groups. We used the difference in median survival time and the log-rank test to assess whether or not survival significantly differed. In addition, we calculated the hazard rate ratio of all causes of death by means of a Cox Proportional Hazard model. The proportionality assumption was tested with a log( log[survival]) plot. All analyses were stratified for histological tumor size (<10 mm, mm, mm, mm, mm, and 50þ mm) according to the classification put forward by Tabar et al. 10 Distinct from other studies looking at long-term survival perspectives, 4,5 our study compares screendetected patients with apparently healthy women. Leadtime bias, that is, overestimation of survival duration among screen-detected patients, is a nonissue in our study, because this bias results from comparison with women otherwise diagnosed with breast cancer, such as nonparticipants or women with interval cancer. Another form of bias that hampers survival analysis is length-time bias. Length-time bias constitutes the overestimation of survival by the relative excess of slowly growing tumors detected at screening. It seems likely that tumors with a long preclinical phase will be much more common at the initial screening examination. Subanalysis for screen-detected patients with repeated screening examinations, especially with a previous negative examination, cancels out the effect of length-time bias. Furthermore, we assume that women who regularly attend screening could have a more beneficial effect of the screening-therapy combination. We defined regular attendees as women with repeated screening examinations Cancer February 1,

3 Original Article who attended at least the last screening examination 2 years before detection and had a negative result. Data on all previous screening examination showed that 90% of these patients accepted >80% of their invitations. All data analyses were performed using the SAS 8.2 package (SAS Institute, Cary, NC). RESULTS Between 1975 and 2006, 885 patients with an average age at diagnosis of 62 years (range, 38-93) were detected with invasive breast cancer in the Nijmegen screening program. At the end of 2007, the number of patients who died of all causes was 343, and the median follow-up time was 8.2 years (range, years). Other patient and tumor characteristics are listed in Table 1. Most of the patients were diagnosed between the ages of 50 and 69 years, and the year of detection was almost equally distributed over the time periods, with an average of 27 screen-detected women per year in , 24 in , and 30 in Regarding screening history, 20% of the cancers were detected at the initial (first) screening examination, whereas 80% of the patients were detected at a repeat screening examination. Most of the women with a repeat screening examination (587 of 686, 86%) had a negative screening result at the previous examination 2 years earlier and were regular attendees (see Materials and Methods). About 60% (n ¼ 515) of all invasive carcinomas concerned a stage I tumor (pt 20 mm and negative lymph nodes) and 39% (n ¼ 337) a stage IIþ tumor (pt >20 mm and/or positive lymph nodes). Negative lymph nodes were found in 62% of all patients with invasive tumors, 27% had positive lymph nodes, and for 11% lymph node status was unknown. The histological tumor size shows that about 2 = 3 of the tumors were <2 cm, whereas size was unknown in 4%. Table 2 shows the life expectancy (median survival) for the screen-detected patients with an invasive tumor and their comparison group and the difference between these 2 groups. The total group of patients (median age, 62.4 years) has a life expectancy of 17.4 years, which is about 6 years shorter than the life expectancy (23.5 years) for the comparison group. The log-rank test showed that the survival curves, as depicted in Figure 1, differed significantly from each other. The differences in life expectancy between the patient and comparison groups vary with tumor size. For the 2 smaller tumor size groups (<10 mm and 10- Table 1. Patient and Tumor Characteristics of 858 Screen- Detected Invasive Breast Cancer Patients ( ) Characteristic No. (%) No. of Deceased up to 2007 All screen-detected patients 858 (100) 314 Age at detection, y <50 85 (10) (70) (20) 96 Period (31) (27) (42) 29 Screening history a Initial screening 172 (20) 92 Repeat screening 686 (80) 222 Previous negative Otherwise Stage b I 515 (60) 169 II 257 (30) 94 III 66 (8) 38 IV 14 (2) 11 Unknown 6 (.7) 2 Lymph node status Negative 536 (62) 156 Positive 230 (27) 99 Unknown 92 (11) 59 Tumor size (pt) <10 mm 171 (20) mm 189 (22) mm 194 (23) mm 170 (20) mm 73 (9) mm 31 (4) 16 Unknown 30 (4) 18 a Initial or repeat screening: detection carcinoma at a woman s first or subsequent screening examination. Previous negative: negative screening result at previous screening examination 2 years earlier. b Four categories of TNM were defined based on surgery; stage I: T1N0M0; stage II: T2N0M0, T1-2N1M0, T3N0M0; stage 3: T3N1M0, T1-3N2M0, T4M0; stage IV: M mm), the survival curves were similar (log-rank test, P ¼.72 and P ¼.86, respectively). The survival curves for the patients with larger tumor sizes differed significantly from the comparison group. Life expectancy was 6 years lower in the tumor size mm group, 10 years lower in the mm group, and 12 years lower in the mm group. This deteriorating trend was not continued in the highest tumor size group of 50þ mm (comprising only 4% of all invasive tumors), where life expectancy was 7 years lower than the comparison group. 588 Cancer February 1, 2010

4 Screen-Detected Breast Cancer Longevity/Otten et al Table 2. Comparison of Life Expectancy and Overall Mortality (Hazard Ratio) in 858 Patients With Screen-Detected Invasive Carcinomas Compared With Women From the Comparison Group (Age and Calendar a Year Matched) Tumor Size Life Expectancy, y Log-Rank Screen-Detected Comparison Difference Test P Patients Group Hazard Ratio (95% CI) All, invasive < ( ) Neg. prev. scr., n < ( ) Histological tumor size b <10 mm ( ) Neg. prev. scr., n ( ) mm ( ) Neg. prev. scr., n þ ( ) mm ( ) Neg. prev. scr., n ( ) mm ( ) Neg. prev. scr., n ( ) mm < ( ) Neg. prev. scr., n ( ) 501 mm ( ) Neg. prev. scr., n ( ) Neg. prev. scr. indicates negative screening result at previous examination 2 years earlier. a Year of detection of the breast cancer patient. b pt unknown: n ¼ 30. The differences in life expectancy between patients who had a previous negative screen examination 2 years before detection and their comparison groups were not as large as for the whole patient group. For the patients with a negative screen examination, the gap becomes smaller; for example, for size group mm it is 2 years (from 6.3 to 4.2 years) (Table 2). For size groups mm and mm, the gap diminishes with 3 and 5 years, respectively. Figure 2A and B illustrate the survival for the regularly screened patients (defined has having a negative previous screen) in the smaller tumor size groups. No difference in the 2 smallest size groups is seen (together as tumor size <15 mm), but there is a significant difference of 4 years in the tumor size group mm versus the comparison group. The last column of Table 2 shows the hazard of dying in both groups, presented as a ratio. The hazard ratio (HR), for all invasive tumors was 1.6 (95% confidence interval [CI], ), indicating a higher risk of death (all causes) compared with the comparison group. The HR was somewhat lower at 1.4 (95% CI, ) in the case of a previous negative screening examination. Considering only the patients with a previous negative screen, for the smaller tumor size groups the HR was not different from 1,whereasforthelargertumorstheHRwaselevated, ranging from 1.8 (95% CI, ) if mm to 3.0 (95% CI, ) in the tumor size group mm. Figure 1. Survival of screen-detected patients with invasive breast cancer (solid line) versus women from the comparison group (dashed line; age and calendar year matched) is shown. The horizontal line indicates median survival (life expectancy). DISCUSSION Our long-term follow-up study reveals that 42% of all invasive breast cancer patients detected at screening have the same life expectancy as women from the comparison group. The group of invasive tumors represents 85% of all screen-detected patients. If carcinoma in situ would also have been included in our study, about half of all screendetected breast cancer patients have the same life expectancy as the comparison group. For other tumor sizes, 15 mm at diagnosis, screen-detected patients life expectancy Cancer February 1,

5 Original Article Figure 2. (A) Survival of screen-detected patients with a previous negative screening examination and a 0-14-mm large invasive breast cancer (solid line) versus women from the comparison group (dashed line; age and calendar year matched) is shown. (B) Survival of screen-detected patients with a previous negative screening examination and a mm large invasive breast cancer (solid line) versus women from the comparison group (dashed line; age and calendar year matched) is shown. differs from the comparison group. The higher risk of death from all causes probably indicates a surplus of breast cancer mortality for these patients. The difference in life expectancy is modified by screening history. The gap in life expectancy between regular attendees and the comparison group diminished, whereas for the group with no previous screening 2 years before detection the difference in life expectancy was detrimentally influenced. For example, patients with tumor size mm without a previous screen 2 years earlier have a 9 years shorter life expectancy than their comparison group (data not shown), whereas this difference was only 4 years for the regular attendees. Thirty-four percent of the women in the comparison group did not participate in the year that their matched breast cancer patient was detected by screening. Given that the participation rate is lower for the comparison group, selection bias may be anticipated. However, subgroup analysis (data not shown) for screen-detected patients and their comparison group of women with equal participation rates reveals that the results were the same, that is, no difference in life expectancy for women with tumors <15 mm, but a difference in life expectancy in the case of tumor sizes >14 mm. The importance of size as a prognostic factor is clear, as metastatic spread is highly correlated with tumor size That screen-detected patients with small tumor size indeed have a good prognosis seems circular reasoning. But, unlike other studies of long-term survival perspectives, our study compares screen-detected patients with the apparently healthy women invited to screening. In addition, we performed subgroup analyses considering differences in life expectancy in case women had a previous negative screen. As explained previously, this approach prevents underestimation of the difference in life expectancy between patients and the comparison group as a result of the incorporation of more slow-growing tumors (length time bias) in initial screening examinations versus repeat examinations (see Materials and Methods). Therefore, the diminished difference in life expectancy for the group of patients with a previous negative screening examination seems valid and indicates an advantage probably caused by regular attendance. Whether this reflects specific lifestyle components of the regular attendees, that is, a healthy screenee bias, 13 remains unclear. The screening program in Nijmegen started in 1975, ahead of the national program in the Netherlands (started in 1989), and is the only longstanding program with a follow-up of >30 years. Although the Nijmegen program initially invited a much broader age range (35þ years) than the national program, the tumor size distribution seems to correspond very well. For repeat screening examination and age category years, 77% of the screen-detected carcinomas were in situ or invasive 2 cm (pt1), which is in line with results from the Dutch national screening program (on average 75%-80%). 14 For interpretation of our results, the following limitations should be kept in mind. Life expectancy presented in our study pertains to patients diagnosed several decades ago (starting from 1975 until 2006). For women diagnosed today, life expectancy may be more favorable because of improvements in screening mammography and treatment. 590 Cancer February 1, 2010

6 Screen-Detected Breast Cancer Longevity/Otten et al Furthermore, one could expect that (breast) cancer survivors adopt lifestyle changes to improve long-term health. 15 Physical activity, 16 diets high in fruits, vegetables, poultry, and fish, and low intake of red meat 17 are likely to protect against mortality from nonbreast cancer causes. Therefore, besides the effect of early detection and treatment, changes in lifestyle after breast cancer detection could contribute to a smaller gap in life expectancy between patients and the comparison group. Unfortunately, we do not know to what extent patients changed their behavior to a healthier lifestyle after they were detected with breast cancer. Our study has yielded evidence for a group of screen-detected breast cancer patients whose life expectancy is similar to that of the comparison group, representing the general population. Details regarding the specific causes of death for these cases could not be derived from our database, and case fatality or excess mortality should be assessed in a future study. In conclusion, our study gives evidence that breast cancer screening brings about the detection of a large proportion of tumors <15 mm, that is, before the development of metastases, with preservation of the general life expectancy. With increasing tumor sizes, life expectancy becomes worse in comparison with the comparison group, but this gap becomes smaller when women regularly participate in a breast cancer screening program. CONFLICT OF INTEREST DISCLOSURES The authors made no disclosures. REFERENCES 1. Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics, CA Cancer J Clin. 2005;55: Shapiro S, Coleman EA, Broeders M, et al. Breast cancer screening programmes in 22 countries: current policies, administration and guidelines. International Breast Cancer Screening Network (IBSN) and the European Network of Pilot Projects for Breast Cancer Screening. Int J Epidemiol. 1998;27: Otten JD, Broeders MJ, Fracheboud J, Otto SJ, de Koning HJ, Verbeek AL. Impressive time-related influence of the Dutch screening programme on breast cancer incidence and mortality, Int J Cancer. 2008; 123: Shen Y, Yang Y, Inoue LY, Munsell MF, Miller AB, Berry DA. Role of detection method in predicting breast cancer survival: analysis of randomized screening trials. J Natl Cancer Inst. 2005;97: Ernst MF, Voogd AC, Coebergh JW, Roukema JA. Breast carcinoma diagnosis, treatment, and prognosis before and after the introduction of mass mammographic screening. Cancer. 2004;100: Bundred NJ. Prognostic and predictive factors in breast cancer. Cancer Treat Rev. 2001;27: Rampaul RS, Pinder SE, Elston CW, Ellis IO. Prognostic and predictive factors in primary breast cancer and their role in patient management: The Nottingham Breast Team. Eur J Surg Oncol. 2001;27: Otten JD, van Dijck JA, Peer PG, et al. Long term breast cancer screening in Nijmegen, the Netherlands: the nine rounds from J Epidemiol Community Health. 1996;50: Duffy SW, Tabar L, Vitak B, et al. The relative contributions of screen-detected in situ and invasive breast carcinomas in reducing mortality from the disease. Eur J Cancer. 2003;39: Tabar L, Fagerberg G, Day NE, Duffy SW, Kitchin RM. Breast cancer treatment and natural history: new insights from results of screening. Lancet. 1992;339: Carter CL, Allen C, Henson DE. Relation of tumor size, lymph node status, and survival in 24,740 breast cancer cases. Cancer. 1989;63: Tubiana M, Koscielny S. The rationale for early diagnosis of cancer the example of breast cancer. Acta Oncol. 1999; 38: Friedman DR, Dubin N. Case-control evaluation of breast cancer screening efficacy. Am J Epidemiol. 1991;133: National Evaluation Team for Breast Cancer Screening. National Evaluation of Mass Screening for Breast Cancer in the Netherlands. Evaluation Report XI [in Dutch with English summary]. Rotterdam, the Netherlands: Department of Public Health, Erasmus MC, University Medical Center Rotterdam; Demark-Wahnefried W, Aziz NM, Rowland JH, Pinto BM. Riding the crest of the teachable moment: promoting long-term health after the diagnosis of cancer. J Clin Oncol. 2005;23: Holmes MD, Chen WY, Feskanich D, Kroenke CH, Colditz GA. Physical activity and survival after breast cancer diagnosis. JAMA. 2005;293: Kroenke CH, Fung TT, Hu FB, Holmes MD. Dietary patterns and survival after breast cancer diagnosis. J Clin Oncol. 2005;23: UICC TNM Classification of Malignant Tumors. 6th ed. New York, NY: Wiley-Liss; 2002: Cancer February 1,