Are Patients Diagnosed With Breast Cancer Before Age 50 Years Ever Cured? Hermann Brenner and Timo Hakulinen

Transcription

1 VOLUME 22 NUMBER 3 FEBRUARY JOURNAL OF CLINICAL ONCOLOGY O R I G I N A L R E P O R T Are Patients Diagnosed With Breast Cancer Before Age 50 Years Ever Cured? Hermann Brenner and Timo Hakulinen From the Department of Epidemiology, German Centre for Research on Aging, Heidelberg, Germany; Finnish Cancer Registry; and Department of Public Health, University of Helsinki, Helsinki, Finland. Submitted April 8, 2003; accepted October 15, T.H. was supported in part by the Ma- DaMe project of the Academy of Finland, Helsinki. Authors disclosures of potential conflicts of interest are found at the end of this article. Address reprint requests to Hermann Brenner, MD, MPH, Department of Epidemiology, German Centre for Research on Aging, Bergheimer Str 20, D Heidelberg, Germany; Brenner@dzfa-uni-heidelberg.de by American Society of Clinical Oncology X/04/ /$20.00 DOI: /JCO A B S T R A C T Purpose Breast cancer diagnosed before the age of 50 years has become a common disease in many developed countries. Although average remaining life expectancy in the affected age groups is usually several decades, data regarding survival perspectives beyond 10 to 20 years after diagnosis are sparse. The aim of this study was to assess long-term survival in a large population-based sample of patients diagnosed with breast cancer before age 50 years. Patients and Methods Relative survival within up to 40 years after diagnosis was assessed for cohorts of women diagnosed with breast cancer before age 50 years and notified to the nationwide Finnish Cancer Registry within various time intervals since In addition, up-to-date estimates of 40-year relative survival were obtained by exclusively looking at the survival experience of breast cancer patients in recent years (1993 to 1999) using period analysis, a new method of survival analysis. Results Prognosis of patients diagnosed with breast cancer before age 50 years has considerably improved during the past decades. According to the latest estimates from period analysis, cumulative 40-year relative survival is now approximately 43% for all cancers combined, 57% for localized cancers, and 24% for cancers with regional tumor spread. Nevertheless, patients diagnosed with breast cancer before age 50 years continue to have increased mortality throughout at least four decades after diagnosis. This applies even if breast cancer is diagnosed in a localized stage and in the absence of a second primary breast cancer. Conclusion Despite major improvement in prognosis over time, breast cancer occurring among patients who are younger than 50 years remains a chronic disease that affects prognosis for decades. J Clin Oncol 22: by American Society of Clinical Oncology INTRODUCTION Breast cancer is the most common cancer among women in developed countries [1], and approximately one of four patients are confronted with this diagnosis before the age of 50 years, in other words, at an age when average remaining life expectancy encompasses several decades. Survival perspectives beyond the usual 5- or 10-year follow-up periods included in most pertinent analyses are therefore of major interest to these patients, their relatives, and clinicians. A few studies have addressed survival rates up to 20 years after diagnosis [2], but data beyond that follow-up interval are sparse. A study from Norway including 458 young breast cancer patients found that an excess breast cancer mortality persisted for at least 40 years after diagnosis [3]. By contrast, mortality was quite similar or even lower among breast cancer patients than in the general population of the same age in the third and fourth decade after diagnosis in studies from New Zealand [4], the Netherlands [5,6], and Finland [7]. However, in most of these studies, numbers of patients for whom long-term follow-up was available were small, and with the exception of the studies from Norway and New Zealand [3,4], specific results for younger breast cancer patients, for whom long-term survival perspectives are of particular concern, have 432

2 Prognosis of Young Patients With Breast Cancer not been reported. Furthermore, previous estimates, which were based on traditional methods of survival analysis, pertained to survival expectations of patients diagnosed several decades ago, and they may therefore be of limited relevance for patients diagnosed today given the major improvement in prognosis of breast cancer that has been achieved within the last few decades [8-12]. In this large population-based study from Finland, we applied a new method of survival analysis to provide up-todate estimates of survival up to 40 years after diagnosis among women diagnosed with breast cancer before the age of 50. PATIENTS AND METHODS Our analysis is based on data from the nationwide Finnish Cancer Registry (population base approximately 5.3 million people), which are among the highest-quality data of any population-based cancer registry in the world. Virtually complete population-based cancer registration has been accomplished since 1953 [13]. Notification of cancer cases to the registry is mandatory by law. Notification comes from many different sources, including hospitals, physicians working outside hospitals, dentists, and pathologic and cytologic laboratories. Copies are also obtained of all death certificates where cancer is mentioned. Mortality follow-up is extremely efficient in Finland because of the existence of personal identification numbers [14]. Using these numbers as the key, the cancer registry files are matched annually with the annual lists of deaths. Matching with the central population registry (a register of all people currently alive and living in Finland) is performed as an additional check on the vital status of patients. By the time of this analysis, follow-up with respect to vital status had been completed until the end of This analysis includes patients diagnosed with breast cancer before age 50 years between 1953 and Patients whose cancer was registered by death certificate or autopsy only (0.15% of registered cases) or whose month of death was unknown (0.2%) were excluded. Forty-year survival curves (overall and according to stage at diagnosis) were first estimated for the cohort of patients diagnosed in 1953 through 1959 and followed over 40 years since then by traditional (cohort-wise) life-table analysis. In addition, 30-, 20-, and 10-year survival curves were derived in the same way for patients diagnosed in 1963 through 1969, 1973 through 1979, and 1983 through 1989, respectively. Next, 40-year survival curves were derived for calendar years 1993 through 1999 by a new method of survival analysis, denoted period analysis [15,16], to obtain more up-to-date estimates of long-term prognosis. The methodologic principle of period analysis, which has been described in detail elsewhere [15,16], is simple. With this approach, the estimates of survival curves are exclusively based on survival experience of patients in some recent period (here, in 1993 through 1999, the most recent period for which data were available). This is achieved by left truncation of observations at the beginning of that period (here, at the beginning of 1993) in addition to right censoring of observations at its end (here, at the end of 1999). The period estimates of cancer patient survival for the period of interest (here, for the 1993 through 1999 period) thus reflect the cumulative survival rates to be expected by newly diagnosed cancer patients, assuming that the conditional survival probabilities within defined time intervals after diagnosis observed in that period prevail. Obviously, these conditional survival probabilities refer to patients diagnosed in various calendar years. For example, in the period analysis for the 1993 through 1999 period, survival experience during the first year after diagnosis is provided by patients diagnosed between 1992 and 1999, survival experience in the second year after diagnosis is provided by patients diagnosed in 1991 through 1998, and so on, until survival experience in the 40th year after diagnosis, which is provided by patients diagnosed in 1953 through These conditional survival probabilities within single years are then multiplied to calculate cumulative survival rates. It has been shown by extensive empirical evaluation that estimates of long-term survival obtained by period analysis for a certain time period quite closely predict the long-term survival of patients diagnosed in that period (which can be obtained by traditional survival analysis only many years later). However, even the most recent period estimates may be somewhat too pessimistic with respect to the long-term prognosis to be expected by recently diagnosed patients in case of ongoing improvement in prognosis, albeit much less so than traditional cohort survival estimates [17-19]. At higher ages, decades after diagnosis, patients diagnosed with breast cancer before age 50 years obviously would be expected to have rapidly increasing mortality, even without having had breast cancer. Therefore, most survival rates presented in this article are relative rather than absolute (observed) survival rates. Relative survival rates reflect so-called net survival of patients with cancer. They can be interpreted as the expected survival experience of cancer patients in the hypothetical situation in which cancer is the only cause of death [20]. Relative survival rates are calculated as the ratio of absolute survival rates of cancer patients divided by the expected survival rates of a group of persons of the corresponding age and sex in the general population, which were obtained from calendar-year specific population life-tables (which include patients with cancer). Estimates of expected survival were derived according to Hakulinen s method [21]. SEs of survival rates were calculated according to Greenwood s method [22]. All analyses were performed with the SAS software package (SAS Institute, Cary, NC) using a publicly available macro for both cohort and period analysis, which is described in detail elsewhere [23]. RESULTS Overall, 18,578 patients diagnosed with breast cancer before age 50 years were included in the analysis. Median age at diagnosis was 44 years (interquartile range, 40 to 47 years). Numbers of patients still alive and under observation at various points of follow-up are listed in Table 1 for the various cohorts of patients for which survival curves are shown. In addition, the corresponding numbers for the 1993 through 1999 period are given. For that period, the numbers of patients under observation at 0, 10, 20, 30, and 40 years after diagnosis equal the corresponding numbers of the cohorts of patients diagnosed in 1993 through 1999, 1983 through 1989, 1973 through 1979, 1963 through 1969, and 1953 through 1959, respectively. The numbers of patients diagnosed with breast cancer before age 50 increased more than three-fold between 1953 through 1959 and 1993 through Although the numbers of patients under 433

3 Brenner and Hakulinen Table 1. Numbers of Patients Under Observation at Various Time Points During Long-Term Follow-Up Included in the Analysis Years After Diagnosis cohort 1, cohort 1, cohort 2,336 1,412 1, cohort 3,513 2, cohort 4,963 2,324 1, observation decrease with increasing length of follow-up, the numbers are sufficiently large to derive reasonably precise estimates throughout the 40 years of follow-up addressed in our analysis. Figures 1 and 2 show the absolute and relative cohort survival curves for patients diagnosed in 1953 through 1959, 1963 through 1969, 1973 through 1979, and 1983 through 1989 as well as the absolute and relative 40-year period survival curve for the 1993 through 1999 period, respectively. Obviously, the absolute survival rates rapidly decrease in the third and fourth decade after diagnosis, when the remaining patients are reaching higher ages (Fig 1). Nevertheless, the 1993 through 1999 period estimate of 40-year Fig 2. Cohort relative survival curves for patients diagnosed with breast cancer before age 50 years in 1953 through 1959, 1963 through 1969, 1973 through 1979, and 1983 through 1989 (- - -) and period relative survival curve for the 1993 through 1999 period ( ). Finnish Cancer Registry, 1953 to 1999 database. Fig 1. Cohort absolute survival curves for patients diagnosed with breast cancer before age 50 years in 1953 through 1959, 1963 through 1969, 1973 through 1979, and 1983 through 1989 (- - -) and period absolute survival curve for the 1993 through 1999 period ( ). Finnish Cancer Registry, 1953 to 1999 database. survival still is approximately 22%, compared with only approximately 13% estimated for the 1953 through 1959 cohort (Fig 1). Ten-year relative survival was only approximately 50% among patients diagnosed in 1953 through Relative survival further decreased to approximately 40%, 35%, and 30%, respectively, at 20, 30, and 40 years after diagnosis, which indicates that these patients continued to have an excess mortality rate compared with the general population during the entire 40-year follow-up interval. Long-term survival rates gradually improved for cohorts of patients diagnosed in later decades, and the 10-year relative survival rates almost reached 70% for the 1983 to 1989 cohort. The period survival curve for the 1993 through 1999 period suggests that prognosis has further improved in recent years. However, a pronounced continuous decline of the survival curve prevailed over the entire 40-year interval (Fig 2). Tumor spread at diagnosis was known for 93.6% of breast cancer patients. Among these, the majority were diagnosed with localized (53.4%) or regional breast cancer (40.9%). Figures 3 and 4 show 40-year relative survival curves for the 1953 through 1959 cohort and the 1993 through 1999 period among these two groups of patients. No separate curves are shown for patients with distant 434 JOURNAL OF CLINICAL ONCOLOGY

4 Prognosis of Young Patients With Breast Cancer Fig 3. Cohort relative survival curve for patients diagnosed with localized breast cancer before age 50 years in 1953 through 1959 (- - -) and corresponding period relative survival curve for the 1993 through 1999 period ( ). Finnish Cancer Registry, 1953 to 1999 database. Fig 4. Cohort relative survival curve for patients diagnosed with regional breast cancer before age 50 years in 1953 through 1959 (- - -) and corresponding period relative survival curve for the 1993 through 1999 period ( ). Finnish Cancer Registry, 1953 to 1999 database. metastases, because of their small numbers (5.1% overall) and because of the low numbers of long-term survivors in this group. As expected, relative survival rates were much worse for patients with metastases in regional lymph nodes than in patients with localized tumors, but relative survival curves were declining over the entire 40-year follow-up interval even for patients with localized cancer. This pattern was particularly evident in the 1993 through 1999 period survival curves, even though these were considerably higher than the 1953 through 1959 cohort survival curves. Overall, approximately 5% of patients had a second primary breast cancer during follow-up. To assess to what extent long-term survival rates are affected by second primary breast cancers, we performed additional analyses in which all observations were censored at the time of diagnosis of a second primary breast cancer. Although this increased relative survival rates to some extent (for example, for the 1993 through 1999 period, the 40-year relative survival rates of patients with localized and regional breast cancer were increased by approximately 5 and 3 percent units, respectively), considerable excess mortality persisted throughout four decades after diagnosis, even in the absence of second primary breast cancers. Table 2 provides a more direct comparison of 10-year relative survival rates within the four subsequent decades of follow-up. For the second, third, and fourth decade, these 10-year relative survival rates are conditional survival rates; in other words, they pertain to patients still alive at the beginning of the respective decade (in contrast to the survival rates shown in the figures that are unconditional, ie, pertaining to the total number of patients at diagnosis). Although 10-year relative survival strongly increased from approximately 50% during the first decade after diagnosis to much higher levels in the subsequent decades in the cohort of patients diagnosed in the period of 1953 through 1959, they remained less than 90% in each decade and even decreased to slightly more than 80% in the fourth decade. In the 1993 through 1999 period analysis, 10-year relative survival was considerably higher in the first decade after diagnosis than for the 1953 through 1959 cohort (71.5% v 48.9%), but similar estimates of 10-year relative survival between 80% and 90% for subsequent decades (with estimates close to 80% for the fourth decade) were obtained in both types of analysis. Among patients with localized breast cancer, the 1993 through 1999 period estimates of 10-year relative survival were quite similar (at levels between 80% and 90%) for all four decades of follow-up; that is, the excess mortality within the next 10 years of patients who had already survived 10, 20, or 30 years was essentially the same as in the first 10 years after diagnosis. Similar patterns were seen for 435

5 Brenner and Hakulinen Table 2. Ten-Year Relative Survival Within Four Decades After Diagnosis (conditional on being alive at the beginning of the decade) for the 1953 Through 1959 Cohort and 1993 Through 1999 Period Years After Diagnosis Relative Survival (%) SE Relative Survival (%) SE Relative Survival (%) SE Relative Survival (%) SE cohort Total Localized Regional period Total Localized Regional the 1953 through 1959 cohort estimates, with the exception of the first 10 years after diagnosis, when relative survival was less than 70%. For patients with regional tumor spread, 10-year relative survival rates increased from much lower levels in the first decade after diagnosis to levels of approximately 80% in the second and third decade after diagnosis, and they decreased to levels of approximately 65% again in the fourth decade after diagnosis. This pattern was seen in both the cohort analysis for the 1953 through 1959 cohort and the period analysis for the 1993 through 1999 period, although 10-year survival during the first 10 years was again considerably lower for the 1953 through 1959 cohort than for the 1993 through 1999 period. The similarity of the cohort and period estimates for the fourth decade is not surprising, given that they are derived from partly overlapping databases. Although the SEs of survival estimates increase from decade to decade during follow-up given the reduction of the number of patients at risk over time, each of the relative survival estimates is by more than two SEs lower than 100%, and most relative survival estimates are by more than four SEs lower than 100%, which indicates that the persistence of excess mortality compared with that of the general population of the same age is highly statistically significant for each of the four decades after diagnosis. DISCUSSION This analysis illustrates that prognosis of patients diagnosed with breast cancer before age 50 years has considerably improved during the past decades. Nevertheless, these patients continue to have increased mortality throughout at least 40 years after diagnosis; in other words, breast cancer diagnosed before the age of 50 years remains a chronic disease that affects prognosis of patients for decades. This applies even if breast cancer is diagnosed in a localized stage, in which case 10-year relative survival rates of between 80% and 90% are now observed for each of the first four decades after diagnosis (conditional on being alive at the beginning of the decade) and in the absence of the occurrence of a second primary breast cancer. For patients with regional tumor spread, excess mortality owing to breast cancer remains substantially higher, particularly during the first and the fourth decade after diagnosis. According to the latest estimates from period analysis, cumulative 40-year relative survival is now approximately 43% for all cancers combined, 57% for localized cancers, and 24% for cancers with regional tumor spread. To our knowledge, this is the largest systematic analysis of long-term survival perspectives of patients diagnosed with breast cancer before age 50 years ever reported, and it is the first one to use period analysis, a new method of survival analysis that allows for more up-to-date estimates of longterm survival perspectives than traditional methods of survival analysis [15-19]. The few previous studies addressing breast cancer survival for more than 20 years after diagnosis were much smaller in terms of numbers of patients, and most had included breast cancer patients at different ages, of whom only a minority were younger than 50 years at the time of diagnosis. Some of these studies had suggested that excess risk of death may cease 20 years after diagnosis [5-7]. By contrast, our results, which are consistent with results from a smaller study specifically addressing long-term perspectives of young breast cancer patients [3], clearly indicate persistence of excess risk for at least four decades among breast cancer patients who are diagnosed before age 50 years. Apart from the fact that very-long-term survival rates are of particular relevance for patients diagnosed at comparably young ages, there are also other reasons to specifically address long-term survival expectations for these patients. Numerous epidemiologic studies have suggested major differences in the etiology and clinical course according to age at diagnosis, and it would not be surprising to see major differences in long-term prognosis as well. In fact, most studies that focused on survival of breast cancer patients 436 JOURNAL OF CLINICAL ONCOLOGY

6 Prognosis of Young Patients With Breast Cancer within 5 to 20 years after diagnosis found prognosis to be better among patients diagnosed with breast cancer in their fifth decade of life than among both younger and older patients [24-27]. Interestingly, in a study from New Zealand, in which very-long-term survival of breast cancer patients at various ages was assessed, patients between 35 and 55 years of age also had the lowest excess mortality within the first 5 years after diagnosis. At the same time, in contrast with older patients, an excess mortality beyond 20 years after diagnosis was observed in these age groups [4]. These findings underline the need to look at very-long-term survival rates to get the full picture of prognosis among patients diagnosed with breast cancer at comparably young ages. In the interpretation of our analysis, the following limitations should be kept in mind. Despite the large overall number of patients involved, further stratification of verylong-term survival estimates by additional factors, such as certain age groups within the limited age range included in our study, was prohibitive because of sample size limitations. Our study also does not provide (and was not intended to provide) evidence regarding potential differences in very-long-term survival rates between patients younger than and older than 50 years, which need to be addressed in additional analyses. Furthermore, details regarding the causes of the excess deaths among patients could not be derived from this registry-based study, but they should be assessed in future work specifically addressing this issue, because their identification might be helpful for focusing long-term clinical surveillance. Our data suggest, however, that second primary breast cancers account for a relatively small proportion of long-term excess mortality. Even though the period method provides more up-todate survival statistics than traditional cohort analysis, it is still using patients diagnosed many years ago to estimate long-term survival. In particular, even the 1993 through 1999 period estimates of 40-year survival were based on estimates of conditional survival in the third and fourth decade after diagnosis of patients diagnosed before screening or modern forms of therapy were widely available (or recommended for women younger than 50 years). Hence, although the 1993 through 1999 period estimates of 40-year survival are much more optimistic and up-to-date than the most recent cohort estimates of 40-year survival, they are likely to underestimate to some extent the survival expectations of patients diagnosed with breast cancer in recent years. Notwithstanding the need to take this issue into account, our results of the period analysis for the 1993 through 1999 period have important clinical implications: according to currently observed survival rates (which are, as far as pertinent data are available, quite similar to those observed in other developed countries, including the United States [28]), breast cancer among patients younger than 50 years is a truly chronic disease, despite the major progress in survival rates in the past decades. For example, according to our latest data from the 1993 through 1999 period, patients with localized breast cancer, who had had their diagnosis made up to 30 years ago, still have the same excess mortality as newly diagnosed patients (Table 2). Nevertheless, our results should not be used to put an additional burden of anxiety on patients diagnosed with breast cancer at young ages. Rather, these results underline the need for state-of-the-art long-term care, as well as further research regarding measures to reduce the long-term excess mortality in this group of patients. Authors Disclosures of Potential Conflicts of Interest The authors indicated no potential conflicts of interest. REFERENCES 1. Parkin DM: Global cancer statistics in the year Lancet Oncol 2: , Sutherland CM, Mather FJ: Long-term survival and prognostic factors in patients with regional breast cancer (skin, muscle and/or chest wall attachment). Cancer 55: , Rutqvist LE, Wallgren A: Long-term survival of 458 young breast cancer patients. Cancer 55: , Hibberd AD, Horwood LJ, Wells JE: Long term prognosis of women with breast cancer in New Zealand: Study of survival to 30 years. BMJ 286: , Nab HW, Hop WCJ, Crommelin MA, et al: Changes in long term prognosis of breast cancer in a Dutch cancer registry. BMJ 309:83-86, Louwman WJ, Klokman WJ, Coebergh JWW: Excess mortality from breast cancer 20 years after diagnosis when life expectancy is normal. Br J Cancer 84: , Joensuu H, Toikkanen S: Cured of breast cancer? J Clin Oncol 13:62-69, Adami H-O, Malker B, Rutqvist L-E, et al: Temporal trends in breast cancer survival in Sweden: Significant improvement in 20 years. J Natl Cancer Inst 76: , Nab HW, Hop WCJ, Crommelin MA, et al: Improved prognosis of breast cancer since 1970 in south-eastern Netherlands. Br J Cancer 70: , Chu KC, Tarone RE, Kessler LG, et al: Recent trends in U.S. breast cancer incidence, survival, and mortality rates. J Natl Cancer Inst 88: , Edwards MJ, Gamel JW, Feuer EJ: Improvement in the prognosis of breast cancer from 1965 to J Clin Oncol 16: , Brenner H, Stegmaier C, Ziegler H: Recent improvement in survival of breast cancer patients in Saarland, Germany. Br J Cancer 78: , Teppo L, Pukkala E, Lehtonen M: Data quality and quality control of a population based cancer registry: Experience in Finland. Acta Oncol 33: , Dickman PW, Hakulinen T, Luostarinen T, et al: Survival of cancer patients in Finland Acta Oncol 38:1-103, 1999 (suppl 12) 15. Brenner H, Gefeller O: An alternative approach to monitoring cancer patient survival. Cancer 78: , Brenner H, Gefeller O: Deriving more upto-date estimates of long term patient survival. J Clin Epidemiol 50: , Brenner H, Hakulinen T: Up-to-date long term survival estimates of patients with cancer by period analysis. J Clin Oncol 20: , Brenner H, Hakulinen T: Very-long-term survival rates of patients with cancer. J Clin Oncol 20: , Brenner H, Söderman B, Hakulinen T: Use of period analysis for providing more up-to-date estimates of long-term survival rates: Empirical evaluation among 370, 000 cancer patients in Finland. Int J Epidemiol 31: ,

7 Brenner and Hakulinen 20. Ederer F, Axtell LM, Cutler SJ: The relative survival rate: A statistical methodology. Natl Cancer Inst Monogr 6: , Hakulinen T: Cancer survival corrected for heterogeneity in patient withdrawal. Biometrics 38: , Greenwood M: A Report on the Natural Duration of Cancer. London, United Kingdom, Ministry of Health, His Majesty s Stationery Office, Brenner H, Hakulinen T, Gefeller O: Computational realization of period analysis for monitoring cancer patient survival. Epidemiology 13: , Adami H-O, Malker B, Holmberg L, et al: The relation between survival and age at diagnosis in breast cancer. N Engl J Med 315: , Sant M, Gatta G, Micheli A, et al: Survival and age at diagnosis of breast cancer in a population-based cancer registry. Eur J Cancer 27: , Levi F, Randimbison L, La Vecchia C: Breast cancer survival in relation to sex and age. Oncology 49: , de la Rochefordiere A, Asselan B, Campana F, et al: Age as a prognostic factor in premenopausal breast cancer. Lancet 341: , Brenner H: Long-term survival rates of cancer patients achieved by the end of the 20th century: A period analysis. Lancet 360: , JOURNAL OF CLINICAL ONCOLOGY