Prognosis of screen-detected breast cancers: results of a population based study

Breast cancer is the most common malignancy in women, both in the developed and developing world, with annual incidence rates ranging from 11.8 per 100,000 in eastern China to 86.3 per 100,000 in North America[1]. During recent decades the incidence of breast cancer has increased in Western Europe and the United States, while the mortality decreased[2, 3]. Part of the increase in incidence reflects the success of breast cancer screening programs[4]. There is sufficient evidence from randomized trials that inviting women 50–69 years of age to be screened with mammography will eventually reduce their mortality from breast cancer. The first randomized clinical trial performed in New York during the 1960s evaluated the value of "early detection" by physical examination and mammography[5], while the Swedish trials, during the 1970s and 1980s (Malmo, Two-County, Stockholm and Goteborg) used only mammography to find out the independent value of this modality [6–9]. A recent review performed by a committee confirmed their previous results [10]. Even another randomized trial in the United Kingdom (Edinburgh) gave the same results[11]. Mammography screening reduces breast cancer mortality of invited women by 20% to 30% in comparison with controls. Mammography screening makes it possible to detect breast cancer at a subclinical or non-palpable stage. Early detection of such small tumours by screening shows a positive effect on survival because of overall decreased mortality from breast cancer [12–14]. Even after a follow-up period as long as 17 years the expected rate of mortality from breast cancer is reduced[15]. In Modena, Italy, a population-based mammography screening for breast cancer was begun in 1995, providing a biennial screening to women aged 50 to 69 years. However, since the effect on breast cancer mortality will only become evident in the long term, monitoring the early outcomes seemed to us essential in the early phases for early evaluation of the efficacy of the screening program. In fact, several studies have shown that the interval from initial diagnosis to recurrence is a relevant prognostic factor [16, 17], since patients with early recurrence tend to do worse than those with late relapse [18].

In the current study we report the results of an exhaustive review of all breast cancers diagnosed between October 1995 and December 2000 in the Province of Modena, in women 50–69 years of age, assessing the rates of small tumors, the patterns of care and the relapse-free, event-free and 5 years survivals in women exposed to screening and in women diagnosed over the same time-interval out of screening.

The present study has shown that breast cancers detected through MSP are usually diagnosed at an early stage and are associated with a very favourable behaviour of the disease and an excellent prognosis, in terms of 5-year event-free survival and Overall Survival. Moreover, the study has also suggested that a population-based cancer registry is a relevant tool in the evaluation of a cancer screening program.

The decision to offer regular mammographic screening is based on the demonstration of reduced breast cancer-specific mortality in prospective randomized clinical trials [21–24]. In the randomized trials of breast screening, comparison of the prognostic profile of cancers detected in the invited as compared with the control groups has identified surrogate intermediate endpoints associated with the mortality reductions in those trials. These endpoints include such factors as the proportion of cancers smaller than 15 mm diameter or absence of lymph node involvement at diagnosis. It has been suggested that these intermediate endpoints might be useful to predict the likelihood that an organized breast screening program will achieve its mandate of mortality reduction on a population basis [25]. The prognostic profile of screen-detected breast cancer within the MSP not only meets the targets derived from the randomized trials, but also demonstrates a significant advantage in terms of initial therapy and event-free survival when compared to other breast cancers diagnosed in the same geographically-defined population over the same time period.

In fact, since the beginning of the organized mammography screening program, an increasing proportion of women with newly diagnosed breast cancer each year (76 in 1996, 100 in 1997, 131 in 1998, 173 in 1999, 175 in 2000) have been MSP participants. The cancers diagnosed in women within four years of the screening program were histologically more favourable (GI-II, hormonal receptors positive, low proliferative activity), smaller (13 mm vs 19 mm) and less likely to have spread to axillary lymph nodes compared to breast cancer diagnosed out of the screening over the same time frame.

Since the crude breast cancer-specific mortality rate cannot be used for an immediate evaluation of the impact of screening mammography, we have used Event-Free Survival at 5 years as a measure of success of the MSP. We considered as events local or distant recurrence of breast cancer, second breast cancer, second cancer other than breast, and death from any cause. The rate of recurrences and second breast cancers, but not of other second malignancies or deaths, was significantly lower in the screen-detected group. As a consequence we observed a statistically significant and clinically meaningful 14% improvement in the 5-year event-free survival of women diagnosed by the MSP compared with those diagnosed out of the screening (89% vs. 75%). A difference was also observed in the subset of node negative tumors, where the 5-year survival rates are 95% in the screen-detected group and 91% in the non participants to the MSP. These results correlate with data from a recent study performed by Joensuu, in which a 5-year survival rate of 97% was observed in women with node negative tumors [26]. In addition Joensuu and colleagues found a significant correlation between tumor size and prognosis. In fact, the 20-year survival rate was 93% for tumors of one cm or less, and only 75% for tumors of more than one cm in size. On the basis of that study we can predict that the excellent 5-year survival rate of 94% observed for tumors detected by the MSP (48% of which have a maximum diameter of one centimeter or less) will translate in a long term (i.e. 10 years) survival exceeding 90%.

Welch et al. [27] and Baum [28], state that 5-years survival rate is not sufficient to demonstrate the effectiveness of a screening program This period is considered too brief to by-pass the lead-time bias, that is the earlier detection of a breast cancer, which increases the survival from the time diagnosis to the time of death, without influencing the natural history of the disease. Calculating the 5-year survival rates after adjustment for lead-time and length bias confirmed our finding of a better prognosis of screen-detected cancers. The analysis of mean follow-up duration for every stage showed a superimposable period between the two groups, particularly for stage I BC, which were the most frequently detected cancers in the screen-detected group. Although the MSP seem sufficiently cost-effective, we have to realize that 29% of the tumours were not detected by screening. In fact, the MSP not only missed a small number of interval cancers (n = 34) but also did not cover the cancers occurring in patients excluded for screening because of a recent mammogram outside the MSP (n = 56) and in women who refused to participate in the screening program. Our results confirm that the characteristics of interval cancers are intermediate, between screen and non-screen detected tumors, in terms of size (17.7 mm), nodal status (24% positive), hormonal receptors (70% positive) and high proliferative activity (34%). Overall the sensitivity of our screening program, considered as a proportion of screen-detected cancers in all cancers diagnosed in the target population was 55%. However, the performance of the MSP compares favourably with the recommendations of the European Commission for quality assurance in mammography screening [29], which indicate a good detection rate at the first round when it is more than 3.5 fold of the incidence rate before the screening (2.6%0) and more than 1.5 fold at the subsequent rounds. Finally, we exclude that a class bias could have influenced our results since the number of women belonging to a low social class in the province of Modena according to the Statistic Service is very low (0.83%). Even though there is no direct information on social class status of population invited to the MSP, in an interview performed for a cervical screening program in Modena province, no differences between women belonging to a high or low social class in terms of participation to the study were found. In fact almost all women participated in the study, whereas in many other countries only women from the higher socioeconomic and educational strata are more likely to accept the invitation to screen and to participate in other health-promoting patterns of behaviour [30].

Our data clearly show that the major cause for the decreasing breast cancer mortality is the smaller tumour size, associated with the earlier detection by the MSP. Moreover, the chemotherapy was not retained as an independent prognostic factor, confirming that the magnitude of impact of chemotherapy is very likely less relevant compared to the effect of early detection. Despite that, in our series of screen-detected breast cancers, a very high rate of patients were treated with chemotherapy, even if the most used regimen was CMF.

Finally, we would remark that our study was made possible and facilitated only by the existence of a population-based cancer registry in the area. Through the registry we recognized all new cases of cancers in women aged 50–69 years and achieved information on follow-up status. We consider a cancer registry an important tool for an adequate monitoring of the screening programs.

In conclusion, our data show the effectiveness of the MSP in terms of early diagnosis, more favorable behaviour of the tumor and better prognosis, even if only 65% of the age-eligible female were partecipating in this program in 1996–2000. This analysis has pointed out the need and potential value of substantially increasing recruitment to the MSP to 70% or more, as achieved in other geographic areas.