糖心vlog

Key Points

Question听 Can rare germline genetic variants be used to discriminate between interval and screen-detected breast cancer?

Findings听 In this genetic association study of 4121 patients with cancer and 5631 healthy controls, protein-truncating variants in the 5 major genes for breast cancer, particularly BRCA1/2 and PALB2 variants, were significantly associated with the occurrence of interval cancer. Women with a diagnosis of interval cancer who were carrying variants in any of these 5 genes experienced lower survival rates compared with women with interval cancer without such variants.

Meaning听 The results of this study suggest that screen-detected and interval cancers possess distinct genetic profiles; consequently, these findings may provide insights into the identification of individuals at high risk of developing aggressive forms of breast cancer.

Importance听 Breast cancers (BCs) diagnosed between 2 screening examinations are called interval cancers (ICs), and they have worse clinicopathological characteristics and poorer prognosis than screen-detected cancers (SDCs). However, the association of rare germline genetic variants with IC have not been studied.

Objective听 To evaluate whether rare germline deleterious protein-truncating variants (PTVs) can be applied to discriminate between IC and SDC while considering mammographic density.

Design, Setting, and Participants听 This population-based genetic association study was based on women aged 40 to 76 years who were attending mammographic screening in Sweden. All women with a diagnosis of BC between January 2001 and January 2016 were included, together with age-matched controls. Patients with BC were followed up for survival until 2021. Statistical analysis was performed from September 2021 to December 2022.

Exposure听 Germline PTVs in 34 BC susceptibility genes as analyzed by targeted sequencing.

Main Outcomes and Measures听 Odds ratios (ORs) were used to compare IC with SDC using logistic regression. Hazard ratios were used to investigate BC-specific survival using Cox regression.

Results听 All 4121 patients with BC (IC, n鈥�=鈥�1229; SDC, n鈥�=鈥�2892) were female, with a mean (SD) age of 55.5 (7.1) years. There were 5631 age-matched controls. The PTVs of the ATM, BRCA1, BRCA2, CHEK2, and PALB2 genes were more common in patients with IC compared with SDC (OR, 1.48; 95% CI, 1.06-2.05). This association was primarily influenced by BRCA1/2 and PALB2 variants. A family history of BC together with PTVs of any of these genes synergistically increased the probability of receiving a diagnosis of IC rather than SDC (OR, 3.95; 95% CI, 1.97-7.92). Furthermore, 10-year BC-specific survival revealed that if a patient received a diagnosis of an IC, carriers of PTVs in any of these 5 genes had significantly worse survival compared with patients not carrying any of them (hazard ratio, 2.04; 95% CI, 1.06-3.92). All of these associations were further pronounced in a subset of patients with IC who had a low mammographic density at prior screening examination.

Conclusions and Relevance听 The results of this study may be helpful in future optimizations of screening programs that aim to lower mortality as well as the clinical treatment of patients with BC.

Although breast cancer (BC) incidence has increased during the last decades, the mortality has decreased over the same period.¹ One of the main reasons for the improved survival is the establishment of mammographic screening programs in many high-income countries.² However, approximately 30% of BCs are not detected at screening³ but rather between 2 screening examinations, and are known as interval BC (IC).

ICs are generally diagnosed at later stages, and the patients are more likely to present metastases in local lymph nodes than patients with a diagnosis of screen-detected cancer (SDC).^4,5 In addition, ICs are more likely to be estrogen receptor (ER)鈥搉egative, progesterone receptor (PR)鈥搉egative,^6,7 and triple-negative⁸ compared with SDC. Thus, ICs have been claimed to be associated with a worse prognosis.⁹ Family history of BC within first-degree relatives has been associated with an increased risk of IC,^10-13 and those patients with IC are also more likely than those with SDC to develop other tumors than BC, indicating a shared genetic cause.¹⁴

Higher mammographic density, quantified as percentage density (PD), reduces mammographic screening sensitivity, thereby increasing the likelihood of a subsequent IC, a phenomenon named masking.⁸ An IC could either be missed or masked at a prior screening, or be a fast-growing tumor. Missed or masked tumors are not considered true IC in contrast to a tumor that was not present at a prior screen. Therefore, the availability of mammographic density data makes it possible to study the determinants associated with true IC.

Genetic determinants of BC have been analyzed in a case-control meta-analysis, including more than 113鈥�000 women from the Breast Cancer Association Consortium (BCAC).¹⁵ A total of 34 known or suspected BC susceptibility genes were investigated. Germline protein-truncating variants (PTVs) in 5 genes (ATM, BRCA1, BRCA2, CHEK2, and PALB2) were strongly associated with a risk of BC.¹⁵ PTVs in these 5 genes (hereinafter referred to as the 5 major genes for BC) are used for providing future cancer risks in the multifactorial Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm (BOADICEA) model^16,17 included in the CanRisk tool.¹⁸ However, to our knowledge, the specific association of these genes with IC or SDC has not been studied.

In this article, we aim to identify germline deleterious PTVs associated with IC rather than SDC while considering mammographic density. Furthermore, we aim to investigate if these PTVs are associated with a worse BC-specific survival among patients with IC.

The study population included women from Swedish cohorts Karolinska Mammography Project for Risk Prediction of Breast Cancer (KARMA) and prevalent KARMA (pKARMA). pKARMA consists of BC cases diagnosed in Stockholm between 2001 and 2008.^8,19 KARMA is a large and well-characterized prospective screening cohort derived from population-based mammographic-screening or clinical radiology examinations conducted at 4 Swedish hospitals.²⁰ BCs diagnosed without a prior screening visit or after a normal screening interval had passed (n鈥�=鈥�2938) were excluded from the analyses as our main interest was ICs and SDCs. The final analytical data set combining pKARMA and KARMA comprised 9752 samples (1229 individuals with IC, 2892 individuals with SDC, and 5631 healthy controls; see eFigure 1 in Supplement 1 for a flowchart of the study creation). Basic information about SDC and IC cases from pKARMA and KARMA cohorts showed no differences in age at diagnosis, percentage mammographic density, or menopausal status, among others (eTable 1 in Supplement 1). All women included in the analyses gave written informed consent to the retrieval of data from medical records and national registers, answered a detailed questionnaire on background and lifestyle risk factors, and provided a blood specimen for genetic analysis. Ethical approvals were given by the ethical review board at Karolinska Institutet (Stockholm, Sweden). For more detailed information on the profile from both cohorts, we refer to our previous publications.^14,21-23

The Stockholm mammography screening program started in 1989, and currently all women aged 40 to 74 years are invited to participate every 2 years. The participation rate for mammography screening in this program is high, exceeding 70%. Full details of the organizational and quality aspects of the program have been previously described.²⁴

All patients with BC were assessed for screening history using dates of mammographic screening visits and information about the outcome, as described previously.^8,14 In short, cancers diagnosed at a regularly scheduled mammographic screening were deemed SDC. Conversely, IC was defined as a BC diagnosis made after a negative screen but before the next visit. Women who did not attend the screening mammogram before BC diagnosis or with missing prior screening information were excluded from our analyses.

Mammographic density was measured in the prior screening mammogram (ie, the last negative mammogram preceding BC diagnosis) using STRATUS.²⁵ Two different approaches were used for the comparison between patients with IC and SDC. The first approach included the entire data set adjusting for PD (based on quartiles). The second approach used a subset of IC cases with low PD, excluding patients with a diagnosis of IC (but not SDC) with medium or high mammographic density (PD, >25). This second approach has been previously used in our group¹⁴ to identify true IC cases not diluted by missed or masked tumors.

For retrieving PTV information, blood samples were sequenced at the Centre for Cancer Genetic Epidemiology (University of Cambridge; Cambridge, England). The 34 known or suspected risk gene panels were sequenced using the Hiseq4000 platform, which covered coding region and exon-intron boundaries. The workflow of sequencing experiments, bioinformatic analyses, and the definition of PTVs can be found in the eMethods in mentioned BCAC study, for which the KARMA cohort was part of the collaborative effort.¹⁵ In short, PTVs were defined as variants that disrupt gene function by introducing a stop codon (nonsense mutation), by frameshift insertion/deletions, or through splice site mutations. PTV carriership was defined as having at least 1 PTV in 1 of the 34 BC genes and was analyzed as a binary exposure (13 patients with BC and 7 controls were carriers of 2 PTVs, and they were considered as carriers of 1 PTV, contributing only in the highest-risk category of the 2 PTVs). The frequency of the 34 predisposition PTVs among all BC cases included in our data set is shown in eFigure 2 and eTable 2 in the Supplement 1.

For retrieving information on family history of BC, we used data from the Swedish Multi-Generation Register.²⁶ For 523 patients (13%) with missing information, questionnaires answered by the participants were used. A positive family history of BC was defined as having a mother and/or sister(s) with BC.

Tumor characteristics were retrieved from the National Quality Register for Breast Cancer.²⁷ Basic adjustments for survival analyses included age at diagnosis (<50, 50鈥�59, 鈮�60 years), year of diagnosis (2001-2004, 2005-2008, and 2008 and onwards), study cohort, and PD. Tumor characteristics adjustments included ER status, PR status, ERBB2 (formerly HER2) status, lymph node involvement, grade (well-differentiated, moderately differentiated, poorly differentiated), and tumor size (<20 mm, 鈮�20 mm). Treatment regimen adjustments included adjuvant chemotherapy, adjuvant hormone therapy, and radiotherapy. Date and cause of death was obtained through matching to the Swedish Cause of Death Register, with virtually no missing data (linkage performed on March 15, 2021).

All statistical analyses were performed in R, version 4.2.0 (R Foundation). Statistical significance was set at 笔鈥�<鈥�.05. We used a logistic regression to identify factors that were significantly different between patients with BC and controls and between IC and SDC.

Multivariable Cox proportional hazard regression models were used to estimate hazard ratios (HRs) for 10-year BC-specific survival from R package 鈥渟urvival,鈥� with time since diagnosis as the underlying time scale. Time at risk was calculated from the date of study entry (ie, blood draw, left truncation) until the date of death due to BC, date of death of any cause, or end of follow-up, whichever came first. Kaplan-Meier plots (univariate, log-rank test) were visualized using the ggsurvplot function in the R package 鈥渟urvminer.鈥�

Compared with controls, patients with BC were more likely to carry PTVs in any of the 34 BC genes. PTVs in the 5 major genes for BC (ATM, BRCA1, BRCA2, CHEK2, and PALB2), were all more often found among patients with BC than among controls (Table 1), and the associated values aligned with the ones reported in the meta-analysis from BCAC.¹⁵

Next, we investigated the association with PTVs in the 5 major genes for BC, directly comparing 1229 patients with a diagnosis of IC with 2892 patients with SDC. Patients with IC were more likely to carry PTVs in the 5 major genes (odds ratio [OR], 1.48; 95% CI, 1.06-2.05) compared with SDC (Table 2). This finding was further pronounced in a subset of IC with breasts of low mammographic density (OR, 1.87; 95% CI, 1.22-2.85), in which only true IC was considered (those cancers with PD <25% and therefore with lower probability of being missed at the previous screening). Using a high-dense subset for IC (PD, 鈮�25%) did not show a significant difference compared with SDC (OR, 1.24; 95% CI, 0.63-1.70).

When stratifying the 5 major genes panel into high risk (BRCA1/2 + PALB2) and moderate-high risk (ATM + CHEK2) for BC, we found that patients with IC were more likely than those with SDC to carry PTVs for BRCA1/2 + PALB2 (Table 2). These associations were also seen in women without a strong family history of BC (defined as 2 or more family members with breast or ovarian cancer, or 1 family member with breast and 1 with ovarian cancer; eTable 3 in Supplement 1).

Having a first-degree family member with BC has previously been found to be associated with an increased risk of IC.^10-13 In our data set, relatives with BC were more often found among patients with IC than among those with SDC (OR, 1.25; 95% CI, 1.05-1.48). Interaction analyses showed that the association between PTVs in the 5 major BC genes and risk of IC vs SDC was statistically higher among women with family history of BC, with ORs of 3.95 (95% CI, 1.97-7.92) and 5.17 (95% CI, 2.26-11.85) for the entire data set and true IC subset, respectively (Table 3).

After excluding 488 in situ BC cases, the prognosis of the remaining 3633 women with BC participating in mammographic screening was investigated. Tumor characteristics of the 3633 patients with BC included in this analysis are described in eTable 4 in Supplement 1. In Cox regression models, carriers of PTVs in any of the 5 major genes were associated with worse 10-year BC-specific mortality than noncarriers (HR, 1.89; 95% CI, 1.08-3.30). Importantly, the significant association between PTV carriership and worse survival remained in the subset comprising the 1154 women with a diagnosis of IC (Figure and Table 4). The fully adjusted model (including tumor characteristics and treatment) showed an association of PTVs in the 5 major genes with worse survival when considering all IC (HR, 2.04; 95% CI, 1.06-3.92) and true IC tumors (HR, 2.97; 95% CI, 1.07-8.25; Table 4), respectively.

In this study, we provided 2 clinical take-home messages. First, we showed that germline PTVs in ATM, BRCA1, BRCA2, CHEK2, and PALB2 were associated with an increased risk of IC, and were mostly associated with BRCA1, BRCA2, and PALB2. Women with a family history of BC, in combination with deleterious variants in any of these 5 genes, were 4 times more likely to develop IC compared with SDC. Second, if a patient received a diagnosis of an IC, carriers of PTVs in any of these 5 genes had a significantly worse survival compared with women not carrying any of the pathogenic variants.

It has been shown that PTVs in 5 of the 34 putative BC genes are strongly associated with risk of BC, placing PALB2 at the highest risk category together with BRCA1/2, and PTVs in ATM and CHEK2 in the moderate鈥揾igh-risk category.¹⁵ Our results aligned with these findings; PTVs in the 5 major BC genes were significantly more common in patients compared with controls. When directly comparing IC vs SDC subsets, we found that PTVs in BRCA1/2 + PALB2 genes conferred a 2-fold increased risk for IC compared with SDC. The 2 remaining major genes combined (ATM + CHEK2) were not statistically associated with IC. These results suggest that SDC and IC are distinct entities in underlying genetics and biology. To our knowledge, this is the first report looking into the genetic differences between SDC and IC using the 5 major genes, potentially making advances in our understanding of genetic susceptibility to these distinct breast cancers. We believe that genetic risk discrimination has potential applicability in clinical care, since women with an IC currently do not benefit from BC screening.

PTVs of the 5 major genes for BC are included in the BOADICEA model for estimating the future risks of developing breast or ovarian cancer, together with information on cancer family history and common genetic variants.^16,17,28-30 A family history of BC has previously been found to be associated with an increased risk of IC.¹³ The results of the current study suggest that patients with PTVs in any of the 5 major genes and family history of BC have a 4-fold to 5-fold increased risk of receiving a diagnosis of IC compared with SDC. The increase was approximately 3 times higher than the expected additive effect of both factors, suggesting unidentified genetic determinants associated with IC, thus providing potentially valuable information for identifying women who are at a very high risk for developing an aggressive BC.

In this study, we offered insights into the genetic underpinnings of the adverse prognostic features associated with IC.^8,31 The results found that carriers of PTVs in any of the 5 major genes displayed a higher mortality among women with IC (approximately 2-fold to 3-fold increased mortality) compared with noncarriers, even after adjusting for known tumor characteristics and treatment. Deficiency in the homologous recombination repair pathway, an accurate, error-free DNA repair mechanism in which all of the 5 major genes have a key role, could potentially be associated with accelerated acquisition of mutations beneficial for tumor growth and survival,^32-34 thus resulting in a worse prognosis for those carrying such variants. The association of the PTVs in those genes with IC aligns with the observation that carriers of those mutations are at risk for more aggressive, higher-grade tumors.³⁵ The elevated tumor grade may be associated with accelerated tumor progression, prompting self-detection by the patient during the intervals between screening rounds rather than at the subsequent screening appointment. Nevertheless, because these mutations are rare, they can explain only a small proportion of IC cases. Most affected women likely possess different, presently unidentified risk factors, whether genetic or nongenetic in nature.

Mammographic screening protocols in Sweden have remained unchanged for decades. Special screening regimens exist for women with BRCA1/2 variants or a strong family history of breast/ovarian cancer.^36,37 These regimens include earlier screening, shorter intervals, and additional imaging methods. Genetic counseling is shifting toward broader gene panels. Our study found that PALB2 variants, along with BRCA1/2 variants, were associated with increased breast cancer risk. This emphasizes the need for more frequent screening for these women, aligning with recent European guidelines.³⁸

While our findings might be relevant for genetic counseling of high-risk families, we foresee the main implication for the general population of women attending population-based mammography screening programs. Implementation of personalized risk-based screening programs is foreseeable in the near future. Our study demonstrates that variant testing in healthy women attending screening might help to identify women at high risk to develop IC. In addition, the diminishing cost of variant testing can provide a high health care value of testing all patients with a new diagnosis of IC to identify those with bad prognosis.

A potential limitation of our study could be the awareness of being a BRCA1/2 carrier and the resulting participation in special screening regimens. However, only 2 patients with BC in our study were aware of carrying a BRCA1/2 variant, which represents a small percentage (3.7%) and is unlikely to have affected our results. Sensitivity analyses of the association between BRCA1/2 (and PALB2) variants and breast cancer in women without a strong family history yielded consistent estimates. One weakness is the limited statistical power for prognosis analyses, especially among patients with SDC, due to a small number of events. In the context of mammography screening investigations, this study finds optimal relevance within countries whose screening programs align with the framework in Sweden. Extrapolating to countries with different screening regimens necessitates further investigation. The study鈥檚 major strength lies in the extensive national data available, representing a large sample of the Swedish population and sequenced women. It provides a platform to understand molecular patterns in BC. Additionally, our study considered mammographic density, allowing us to identify aggressive interval tumors that were not missed during screenings but emerged between screens.

The results of this genetic association study suggest that women who inherited PTVs of the 5 major genes for BC disproportionately contributed to the diagnosis of IC, and it was mostly associated with BRCA1/2 and PALB2 genes. The synergistic effect of family history of BC and the 5 major genes suggests that further large-scale sequencing efforts are necessary to uncover the full genetic contribution of the observed interaction. Our results also suggested that PTVs in those 5 major genes are associated with a worse survival among patients with IC. Our work potentially clarifies the picture of what type of BC is likely to evade detection in population-based screening programs. These insights will possibly be helpful in future optimizations of screening programs aimed at lowering mortality, as well as the clinical treatment of patients with BC.

Accepted for Publication: October 16, 2023.

Published Online: January 25, 2024. doi:10.1001/jamaoncol.2023.6287

Open Access: This is an open access article distributed under the terms of the CC-BY License. 漏 2024 Rodriguez J et al. JAMA Oncology.

Corresponding Author: Juan Rodriguez, PhD, Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Nobels v盲g 12A, 171 77 Stockholm, Sweden (juan.rodriguez@ki.se).

Correction: This article was corrected on March 21, 2024, to fix errors in the Key Points and Abstract.

Author Contributions: Drs Rodriguez and Czene had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Rodriguez, Grassmann, Bajalica-Lagercrantz, Hall, Czene.

Acquisition, analysis, or interpretation of data: Rodriguez, Grassmann, Xiao, Eriksson, Mao, Hall, Czene.

Drafting of the manuscript: Rodriguez, Czene.

Critical review of the manuscript for important intellectual content: All authors.

Statistical analysis: Rodriguez, Grassmann, Xiao.

Obtained funding: Hall, Czene.

Administrative, technical, or material support: Rodriguez, Eriksson, Hall.

Supervision: Grassmann, Bajalica-Lagercrantz, Hall, Czene.

Conflict of Interest Disclosures: None reported.

Funding/Support: This work was supported by the Swedish Research Council (grant 2022-00584), the Swedish Cancer Society (grants 22 2207 and 19 0267), and the Stockholm County Council (grant 20200102).

Role of the Funder/Sponsor: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Data Sharing Statement: See Supplement 2.