Fax: +41 21 323 03 03.
Invasive breast cancer following ductal and lobular carcinoma in situ of the breast
Article first published online: 18 APR 2005
Copyright © 2005 Wiley-Liss, Inc.
International Journal of Cancer
Volume 116, Issue 5, pages 820–823, 20 September 2005
How to Cite
Levi, F., Randimbison, L., Te, V.-C. and La Vecchia, C. (2005), Invasive breast cancer following ductal and lobular carcinoma in situ of the breast. Int. J. Cancer, 116: 820–823. doi: 10.1002/ijc.20870
- Issue published online: 20 JUL 2005
- Article first published online: 18 APR 2005
- Manuscript Accepted: 5 OCT 2004
- Manuscript Received: 4 AUG 2004
- breast neoplasms;
- cancer registry;
- ductal carcinoma in situ;
- lobular carcinoma in situ;
- noninvasive breast neoplasms;
- second primary
We considered the risk of subsequent invasive breast cancer in a population-based series of 579 carcinomas in situ (CIS) of the breast (482 ductal, 88 lobular) registered between 1977 and 2002 in the Swiss Canton of Vaud. A total of 55 cases of invasive breast cancer were observed vs. 12.3 expected, corresponding to a standardized incidence ratio (SIR) of 4.5 (95% confidence interval [CI], 3.4–5.8). The SIR was 4.6 after ductal and 4.2 after lobular CIS, was similar with passing time since CIS diagnosis, but was higher (SIR = 5.5) for women aged <55 years. At 20 years following CIS, the cumulative risk of invasive breast cancer was 26%, similar for lobular and for ductal CIS. The incidence of invasive breast cancer following CIS showed no consistent pattern of trends with age, all rates in subsequent age groups ranging between 10 and 18 in 1,000. This is compatible with the occurrence of a single mutational event in a population of susceptible women. © 2005 Wiley-Liss, Inc.
Carcinoma in situ (CIS) of the breast is a preinvasive breast tumor that has become increasingly commonly detected in women undergoing (screening) mammography.1, 2, 3 In recent years, CIS accounted for over 20% of breast cancers diagnosed in the United States (i.e., over 50,000 cases per year)1, 4 and Europe.5, 6 Substantial upward trends were observed earlier for ductal CIS (DCIS)7 and, more recently, for lobular CIS (LCIS).5, 6
The prognosis of women diagnosed with breast CIS is also poorly defined.2 In large clinical trials from the United States8, 9, 10, 11 and Europe,12, 13, 14, 15 the risk of ipsilateral invasive breast cancer varied between 14 and 32% according to length of follow-up and type of therapy, and that of contralateral breast cancer varied between 2 and 6.6%.
Scanty population-based data are available. Among them, in a series of 28 noncomedo DCIS from the United States,16 the risk of invasive breast cancer was elevated up to 30 years after diagnosis. Among 1,929 women with DCIS and 282 with LCIS from western Washington state, the risk of contralateral invasive disease was about 2-fold after DCIS and 3-fold after LCIS.17 In the U.S. Surveillance, Epidemiology and End Results (SEER) Program, 1,525 DCIS in women 40 and older were diagnosed between 1978 and 1983 (before screening mammography was common), and 5,547 between 1984 and 1989 (when screening mammography became common). No information on incidence of subsequent invasive breast cancer was available, but the relative risks (RR) of breast cancer death were 3.4 in the first period and 1.9 in the second one.18 Another series of 3,455 CIS cases from Sweden, including 116 cases of invasive cancer, gave an RR of 4.5 for DCIS and of 4.0 for LCIS.15 A series of 4,198 women diagnosed with CIS between 1975 and 1998 in the Connecticut Tumor Registry showed a 10-year cumulative probability of contralateral breast cancer (invasive and CIS) of 6.8% for women with DCIS and of 13.9% for those with LCIS.19
The follow-up over the period 1977–1994 of 249 breast CIS registered in the Swiss Canton of Vaud showed an RR of breast cancer of 7.2 and a cumulative incidence at 15 years of 24%, based on 24 invasive cancer cases.20 We have now updated that series, to include 579 incident CIS followed for up to 25 years.
Material and methods
Data for the present report were abstracted from the Vaud Cancer Registry file, which includes incident cases of malignant neoplasms in the canton5, 21 whose population, according to the year 2000 census, was about 641,000 inhabitants. A registration scheme applying the same standardized rules as for incident malignancies, has been implemented for CIS of the breast. In particular, all histological reports from pathology laboratories were scrutinized and reviewed when reporting diagnosis of CIS.
The present series comprised 579 women diagnosed between 1977 and 2002 with histologically confirmed CIS. These included 482 DCIS (317 noninfiltrating intraductal carcinomas [ICD-O M: 8500]; 104 noninfiltrating comedocarcinomas [ICD-O M: 8501]; 55 noninfiltrating intraductal papillary carcinomas [ICD-O M: 8503]; 6 noninfiltrating intracystic carcinomas [ICD-O M: 8504]), 88 LCIS (ICD-O M: 8520) and 9 CIS of other or unspecified type.22 The age range was 27–91 years (median age 55 years).
These 579 women with CIS were followed up to the end of 2002 for the occurrence of cancer (excluding cutaneous nonmelanomatous neoplasms), migration or death, corresponding to a total of 4,025 person-years at risk. Passive and active follow-ups were recorded, and each subsequent item of information concerning an already registered case was used to complete the record of the patient. Overall, losses to follow-up did not exceed 3%. A second primary breast cancer was defined as a new malignancy occurring in a patient notified as having had a CIS of the breast, and was classified as independent if so specified by pathological report.20
Cumulative risk was computed using the life table approach. Computation of expected numbers was based on 5-year age group, and calendar year-specific incidence rate multiplied by the observed number of person-years at risk; 95% confidence intervals [CIs] of the observed/expected ratios (standardized incidence ratios [SIR]) were based on the Poisson distribution.
Table I gives the number of observed and expected invasive breast cancers following CIS. Overall, 55 cancers were observed vs. 12.3 expected, corresponding to a SIR of 4.5. Of these, 29 were ipsilateral (25 after DCIS, 4 after LCIS), 24 contralateral (19 after DCIS, 5 after LCIS) and the remaining 2 were invasive left-sided following a bilateral CIS and a CIS of unknown laterality. The SIR was similar with time since CIS diagnosis (SIR = 4.7 for <1 year; 4.7 for 1–4 years, 4.4 for ≥5 years), appeared to decline with increasing age at diagnosis (SIR = 5.5 at age <55 years and 3.6 at age ≥55 years), was 5.2 for CIS diagnosed in 1977–1989 and 3.9 for those diagnosed in 1990–2002 (when screening mammography at age 50–69 years became widespread in this population). Information on screen-detected CIS was available after 1989. Among 277 screen-detected CIS, 13 invasive breast cancers were observed vs. 4.08 expected (SIR = 3.19; 95% CI, 1.7–5.5). The SIR was 4.3 for ductal invasive breast cancer, and 4.8 for lobular. Among the 104 comedo carcinomas, 9 had subsequent invasive breast cancer (SIR = 5.14, 95% CI, 2.3–9.8). None of these estimates was significantly heterogeneous.
|O||E||SIR (95% CI)||O||E||SIR (95% CI)||O||E||SIR (95% CI)|
|Years since diagnosis|
|<1||7||1.5||4.7 (1.9–9.7)||4||1.3||3.1 (0.8–8.0)||3||0.2||15.1 (3.0–44.1)|
|1–4||22||4.7||4.7 (2.9–7.1)||19||3.9||4.8 (2.9–7.6)||3||0.7||4.1 (0.8–12.0)|
|5–9||16||3.5||4.6 (2.6–7.4)||16||2.7||5.9 (3.3–9.5)||0||–||–|
|≥10||10||2.5||3.9 (1.9–7.2)||5||1.6||3.2 (1.0–7.5)||5||1.0||5.2 (1.7–12.2)|
|Age at diagnosis (years)|
|<55||31||5.6||5.5 (3.7–7.8)||24||3.8||6.3 (4.0–9.4)||7||1.7||4.0 (1.6–8.3)|
|≥55||24||6.7||3.6 (2.3–5.4)||20||5.7||3.5 (2.1–5.4)||4||0.9||4.6 (1.2–11.6)|
|Calendar period of diagnosis|
|1977–1989||28||5.4||5.2 (3.4–7.4)||21||3.6||5.9 (3.6–9.0)||7||1.8||4.0 (1.6–8.2)|
|1990–2002||27||6.9||3.9 (2.6–5.7)||23||5.9||3.9 (2.5–5.8)||4||0.9||4.7 (1.3–12.0)|
|Histological type of breast cancer|
|Ductal||38||8.8||4.3 (3.1–5.9)||33||6.7||4.9 (3.4–6.9)||5||1.9||2.6 (0.8–6.1)|
|Lobular||10||2.1||4.8 (2.3–8.7)||5||1.6||3.1 (1.0–7.1)||5||0.4||11.5 (3.7–26.8)|
|Others||7||1.4||5.0 (2.0–10.3)||6||1.1||5.4 (2.0–11.8)||1||0.3||3.6 (0.0–20.1)|
|Total||55||12.3||4.5 (3.4–5.8)||44||9.5||4.6 (3.4–6.2)||11||2.6||4.2 (2.1–7.5)|
Table I also gives corresponding data in separate strata of DCIS (482 cases, 44 observed invasive breast cancers, SIR = 4.6) and LCIS (88 cases, 11 observed invasive breast cancers, SIR = 4.2). The SIR of ductal invasive breast cancer was appreciably higher following DCIS (SIR = 4.9), as was that of lobular breast cancer following LCIS (SIR = 11.5). None of the other strata showed any consistent pattern of risk. A total of 22 other cancers were observed vs. 21.2 expected (SIR = 1.0, 95% CI, 0.7–1.6).
Figure 1 gives the cumulative risk of invasive breast cancer at various time intervals up to 20 years following diagnosis of CIS, at all ages and at age <55 and ≥55 years (Fig. 1a), as well as for DCIS and LCIS (Fig. 1b). The estimated overall cumulative risk after 20 years was 25.9% (26.6% at age <55 and 24.4% at ≥55 years); it was 22.0% for DCIS and 27.5% for LCIS.
Figure 2 gives the age distribution of invasive breast cancer following CIS. All the rates between age 30–39 and ≥80 years ranged between 10 and 18 per 1,000 (average rate, 13.7 per 1,000), in the absence of any consistent pattern of trends across subsequent age groups.
A total of 17 deaths from cancer of the breast were found vs. 2.7 expected. The standardized mortality ratio was 6.4 (95% CI, 3.7–10.2), whereas the cumulative risk of death was 3.3% at 10 and 15.5% at 20 years.
Our work includes one of the largest available population-based datasets of breast CIS, followed for over 20 years for cancer development or death from breast cancer. It therefore quantifies the substantial excess risk not only of breast cancer incidence, but also of mortality, following CIS. The latter is of specific interest since, while information on incidence may be subject to more accurate diagnosis and ascertainment bias as compared to the general population, data on breast cancer deaths in this population are virtually unbiased. The effect of earlier diagnosis of CIS over the last few years is also reflected in the smaller excess risk of invasive breast cancer following CIS diagnosed over recent calendar periods.
Over the last 2 decades, not only diagnosis but also treatment of DCIS has changed. In the U.S. SEER database, the incidence of DCIS dramatically increased between 1992 and 1999, but the rate of mastectomy decreased, leaving the incidence of mastectomy stable. Patients were also less likely to undergo axillary dissection over time.23 The impact of changed therapeutic approach on subsequent breast cancer remains, however, unclear.24
The lack of information on CIS treatments in the present dataset does not allow us to estimate their impact on subsequent invasive breast cancer risk. The data, however, indicate that the RR of invasive breast cancer is approximately constant with passing time following CIS diagnosis, and that the RR is appreciably higher at younger age. Thus, as for contralateral breast cancer, the incidence of invasive breast cancer following CIS was independent from age.25, 26, 27 This pattern is compatible with the occurrence of a single mutational event in a population of susceptible women.28, 29 As expected, ductal invasive breast cancers were more common after DCIS, and lobular after LCIS.
The computation of the expected number of breast cancers in prospective studies in which breasts have partly been removed is, however, likely to be overestimated,30, 31 although any precise quantification of such bias in women with CIS remains unclear. The overall incidence of invasive breast cancer following CIS in the present dataset is around 14 per 1,000, whereas that of contralateral breast cancer following invasive breast cancer in the same population who have only one breast at risk is around 7 per 1,000.27 This indicates that any potential bias is minor.
We thank the staff of the Vaud Cancer Registry for their contributions to this study.
- 21Statistics from the Registry of the Canton of Vaud, Switzerland, 1993-1996. In: ParkinDM, WhelanS, FerlayJ, TeppoL, ThomasDB, eds. Cancer incidence in five continents, vol. VIII, IARC Scientific Publication No 155. Lyon: International Agency for Research on Cancer, 2002. 460–1., , .
- 22World Health Organization. International classification of diseases for oncology, ICD-O. Geneva: World Health Organization, 1976. 131.