Evolution of Breast Cancer Incidence in Young Women in a French Registry From 1990 to 2018

Towards a Change in Screening Strategy?

Yanis Hassaine; Emmanuelle Jacquet; Arnaud Seigneurin; Patricia Delafosse


Breast Cancer Res. 2022;24(87) 

In This Article

Materials and Methods

Study Design

This retrospective observational study was based on data from an ongoing population-based cancer registry in Isère, a French administrative entity with nearly 1.2 million inhabitants. It is a department located in the southeast of France, in a mountainous region where the city of Grenoble is the prefecture.

Study Population

We included women with invasive BC in the Isère department from January 1990 to December 2018.

All first incident female BC was included. Breast sarcomas, lymphomas, and carcinomas in situ were excluded.

Data Collection

Data were collected by the Isère Cancer Registry, which collects incident cancer cases from different sources including histopathology laboratories, oncology departments, social security offices, and medical databases.

The following variables were used: dates of birth and diagnosis as well as cancer site and tumor morphology according to the International Classification of Disease for Oncology, ICD-O.

From 2011, supplementary clinical data were collected for each case: the hormone receptor (estrogen receptor (ER), progesterone receptor (PR)) and the human epidermal growth factor receptor type 2 (HER2) status of the tumor. Therefore, we defined four groups according to the tumor subtype: luminal (RH +), triple negative (RH- and Her2-), Her 2 amplified (Her2 +, RH-), and patients with no information on tumor subtype.

For cases diagnosed between 1990 and 2010, information on tumor subtypes was not available in the registry database.

Statistical Analyses

Annual world standardized incidence rates were calculated for each calendar year from 1990 to 2018. We then computed incidence rates by calendar years among four age groups: < 40 years (young women), 40–49 years, 50–74 years (women invited to organized screening), and 75 years and over. Annual world standardized incidence rates were also computed by tumor subtypes from 2011 to 2018.

Incidence rates for all BC during the 1990–2018 period were modeled using Poisson regression with restricted cubic splines to model age, period, and cohort effects. The number of degrees of freedom for the cubic splines was determined considering the AIC criterion. The average annual percent change (AAPC) of incidence rates was computed using the model coefficients, and its 95% confidence intervals were obtained with the delta method. Separate Poisson regressions were used to model incidence rates for luminal, Her2, and triple-negative BC during the 2011–2018 diagnostic period.

We computed 10-year relative survival rates for each age group for cases diagnosed during the 1990–1999 and during the 2000–2008 periods. Relative survival estimated survival rates associated with mortality from breast cancer by considering time from diagnostic to all-cause death during a 10-year follow-up and by incorporating background mortality, which was obtained from national life tables. Indeed, in population-based studies, the estimation of survival rates using cause of death data can be problematic because the cause of death can be unreliable and treatment-related deaths are not always attributed to the initial disease. Relative survival avoids these problems by considering the total mortality rate as a sum of the expected mortality rate (obtained from national life tables matched on age, sex, and year) and the excess mortality rate associated with BC. Three relative survival models were built for luminal, Her2-amplified, and triple-negative BC. Flexible parametric relative survival models were used, and each model included categorical variables for age at diagnosis (< 40; 40–49; 50–74; ≥ 75) and metastasis at diagnosis.[16] Interaction terms between age groups and metastasis at diagnosis and time-varying effects of age groups and metastasis at diagnosis were also included in the models considering the AIC criterion. We finally computed 5-year relative survival rates by tumor subtypes for cases diagnosed during the 2011–2013 period.

Analyses were realized using Stata 16.1.