Gene expression differences
In 2008, Anders and colleagues [
20] published one of the first attempts to describe the biology of breast cancer in young women using gene expression profiling. In this analysis, which included 200 patients in the young group (≤45 years) and 211 patients in the control group (≥65 years), a higher probability of phosphatidylinositide 3-kinase (PI3K;
P = 0.006) and Myc (
P = 0.03) pathway deregulation was observed in tumors arising in younger patients. However, this analysis was not adjusted for potential differences in breast cancer molecular subtypes as well as other known prognostic factors. Subsequently, a similar analysis was performed by the same group with appropriate adjustment for molecular subtypes among other features [
17], using two publicly available datasets; the first including 48 patients aged ≤45 years and 144 patients ≥65 years, and the second including 92 patients ≤45 years and 108 patients ≥65 years. As expected, younger patients in their datasets had more basal-like tumors but, after adjustment for subtype differences, no distinct molecular aberrations were found that were related to age.
More recently, Azim and colleagues [
9] conducted a pooled gene expression analysis on two datasets including 1,188 and 2,334 patients. The aim was to evaluate the association between patients’ age and nearly 50 genes that were identified based on literature search to be related to early-onset breast cancer. The analysis was adjusted for differences in breast cancer molecular subtype, histological grade, tumor size, and nodal status. Results on the first dataset (n = 1,188, ≤40 years = 191) showed that, independent of subtype, grade and stage, younger patients have higher expression of
RANK-ligand (
P < 0.0001),
c-kit (
P < 0.001), in addition to mammary stem cell (
P < 0.0001) and luminal progenitors and
BRCA1 mutation signatures (
P = 0.007). In addition, there was more disruption of the mitogen-activated protein kinase and PI3K pathways (
P < 0.0001) and lower expression of
BRCA1 (
P = 0.003) and several apoptosis-related genes, particularly
FAS (
P = 0.03). The very same findings were reproduced in an independent dataset that included 2,334 patients, of whom 260 were aged ≤40 years. At a glance these results appear in direct conflict with those of Anders and colleagues [
17]. However, the Anders analysis included four times fewer patients and utilized an unbiased approach in searching for genes associated with age, which requires a relatively high number of patients, especially with adjustment for several confounders and multiple comparisons.
The results by Azim and colleagues suggest interesting insights into the biology of early onset breast cancer. The high
BRCA1 mutation signature expression is consistent with the known relatively high prevalence of
BRCA1 mutations in younger patients [
21],[
22]. Patients with
BRCA1 mutations are commonly diagnosed with basal-like tumors [
23]; earlier work suggested that luminal progenitors appear to be the cell of origin of these tumors and are regulated by c-kit [
24]. The high expression of the
BRCA1 mutation signature, luminal progenitors and c-kit in younger patients all may suggest why young women tend to develop basal-like tumors at higher frequencies. The high expression of RANKL (Receptor activator of nuclear factor kappa-B ligand) is also intriguing given RANKL is known to stimulate osteoclastogenesis and targeting RANKL has been shown to reduce risk of osteoporosis and related skeletal events secondary to bone metastases [
25]. RANKL has also emerged as a PgR-regulated gene that is involved in the expansion of mammary stem cells, increasing their proliferation and protecting them from undergoing apoptosis [
25]. In young women, the normal breast is enriched with an immature mammary cell population (that is, stem cells and progenitors), which increases during pregnancy and breastfeeding, an effect that has been shown to be mainly regulated by RANKL [
26]. In preclinical breast cancer models, RANKL inhibition arrested progestin-induced cancer and reduced the mammary stem cell component [
27]. Thus, RANKL appears to be a potentially relevant breast cancer target beyond its established role in managing bone metastases.
Prognostic genomic signatures in young breast cancer patients
Currently several genomic tests are available to improve prognostication and aid decision making in the adjuvant setting [
16]. This includes Oncotype Dx®, Mammaprint®, Endopredict, PAM50, Breast Cancer Index and many others [
28]. They add prognostic information to classic prognostic variables in patients with ER-positive tumors and appear to distinguish reliably between patients at low and high risk of recurrence [
28]. They are increasingly integrated in standard clinical practice, yet there have been concerns about whether they carry the same prognostic value in young women given these signatures were mainly developed using populations of postmenopausal women.
The initial work by the Dutch group on MammaPrint® including 295 patients, only 63 (21%) of whom were younger than 40 years, revealed 52/63 young patients (82%) were classified as high risk [
29]. The same was observed in earlier studies with Oncotype Dx®, where only 59 out of 668 patients were aged less than 40 years, yet the majority had a high risk score (33/59 young patients; 56%) [
30]. This was somewhat higher than the proportions in the high risk group in patients aged 40 to 50 (29%), 50 to 60 (25%) and >60 years (21%). The other signatures were also largely developed using populations of older patients and hence it is hard to extrapolate from these studies the value of genomic signatures in the young population.
A pooled gene expression analysis recently addressed the prognostic value of three signatures according to age: GENE70 (the microarray version of MammaPrint®), the genomic grade index and GENE76 [
9]. In an analysis including 755 patients with ER-positive disease, of whom 87 were aged ≤40 years, each of the genomic signatures was significantly associated with disease-free survival and added significant prognostic information to the clinical risk classifier, Adjuvant Online. The prognostic value was the same across all age groups, suggesting that genomic signatures can add prognostic information in younger as well as older women with breast cancer.
Pattern of mutations in young breast cancer patients
Several recent studies have reported on the landscape of somatic mutations in breast cancer using next generation sequencing [
31]–[
33]. Point mutations have been observed in
TP53 and
PIK3CA genes, accounting for nearly 25% of cases. However, very little is known regarding the pattern of somatic mutations in younger women. Stephens and colleagues [
33] conducted whole genome sequencing of 100 breast tumors but found no correlation between total number of somatic base substitution and age at diagnosis in both ER-positive (
P = 0.33) and ER-negative (
P = 0.14) tumors.
Recently, the pattern of hot spot somatic mutations using Sequenom was evaluated in 167 young breast cancer patients (mean age of 36 years), of whom 54 were diagnosed during pregnancy [
34]. A total of 84 mutations in 19 genes were evaluated, including 29 different mutations of
PIK3CA (94% of known mutations), and 7 and 6 mutations for
ERBB2 and
TP53, respectively. No differences were observed between the pregnant and non-pregnant groups. While this study lacked a control group, the prevalence of mutations particularly in
PIK3CA appeared to be in line with their known prevalence in older women: approximately 23%. Only 5% of patients had a
TP53 mutation, although it should be noted that only 12% of known
P53 mutations were explored in this study. No
ERBB2 mutations were observed at all.
Regarding germline mutations,
BRCA1/2 mutations are the most common, accounting for up to 40% of familial breast cancer [
35]. In a large analysis including 3,345 patients who were aged ≤50 years at the time of breast cancer diagnosis, 7% of patients had a
BRCA1 mutation [
21]. However,
BRCA1 carriers were significantly younger (mean age 41.9 versus 44.1,
P < 0.001), and had more ER-negative (84.1% versus 38.1%,
P < 0.001) and HER2-negative (93% versus 79%,
P < 0.001) tumors.
Data on other familial breast cancer syndromes in young women are very scarce. CHEK2*1100delC is another germline mutation that has been described to occur more commonly in younger patients. A recent study from Denmark evaluating 25,571 patients found that 1.8% were CHEK2*1100delC heterozygous [
36]. These patients were younger and were more likely to be premenopausal and have ER-positive disease (all
P < 0.001).
The fact that women with familial breast cancer syndromes appear to develop the disease more frequently at an earlier age adds further complexity to the biological make-up of breast cancer in young women. Further research to elucidate the triggers for the development of disease in this high risk young population in particular is clearly warranted.