Invasive Lobular Carcinoma in the Screening Setting

Abstract

Invasive lobular carcinoma (ILC) is the second-most common histologic subtype of breast cancer, constituting 5% to 15% of all breast cancers. It is characterized by an infiltrating growth pattern that may decrease detectability on mammography and US. The use of digital breast tomosynthesis (DBT) improves conspicuity of ILC, and sensitivity is 80% to 88% for ILC. Sensitivity of mammography is lower in dense breasts, and breast tomosynthesis has better sensitivity for ILC in dense breasts compared with digital mammography (DM). Screening US identifies additional ILCs even after DBT, with a supplemental cancer detection rate of 0 to 1.2 ILC per 1000 examinations. Thirteen percent of incremental cancers found by screening US are ILCs. Breast MRI has a sensitivity of 93% for ILC. Abbreviated breast MRI also has high sensitivity but may be limited due to delayed enhancement in ILC. Contrast-enhanced mammography has improved sensitivity for ILC compared with DM, with higher specificity than breast MRI. In summary, supplemental screening modalities increase detection of ILC, with MRI demonstrating the highest sensitivity.

Introduction

Invasive lobular carcinoma (ILC) is the second-most common histologic subtype of breast cancer after invasive ductal carcinoma (IDC) and constitutes 5% to 15% of all breast cancers.¹ Pathologically, cells of ILC are characterized by loss of E-cadherin expression, which results in lack of cell-to-cell cohesion. The growth pattern can be one of infiltrating single cells or a single-file line of cells, sheets of cells, or aggregates of small numbers of cells.² Due to these diffuse growth patterns, ILC is more difficult to detect by physical examination or by imaging compared with other histologic types of breast cancer.³ Invasive lobular carcinoma is also more likely to be multifocal/multicentric (36% of ILCs are multifocal or multicentric compared with 11% of IDCs) and bilateral compared with IDC (20.9% of ILCs are bilateral compared with 11.2% of IDCs).^4-6 It is also more likely to be diagnosed at a higher stage compared with IDC, with larger tumor size (12.4% of ILCs present as T3 or T4 compared with 6.4% for IDC), with higher likelihood of nodal metastases (59.1% of ILCs present without nodal metastases compared with 62% for IDCs), and with higher likelihood of distant metastatic disease (5.5% of ILCs present with metastatic disease compared with 3.8% for IDCs).⁷

While approximately 5% of invasive breast cancers have mixed features of both ductal and lobular subtypes, these mixed tumors are a poorly understood subtype of invasive breast cancer and share clinical features of both IDC and ILC.^2,8 Imaging research on this histologic subtype is sparse, although a single report indicates that the imaging characteristics are more similar to ILC than IDC.⁹ In this review, we exclude mixed ductal-lobular carcinoma and focus on pure ILC.

The purpose of this article is to review the evidence in screening detection of ILC across modalities. Studies of mammography, digital breast tomosynthesis (DBT), US, MRI, and contrast-enhanced mammography (CEM) that break down detected cancers by histology are included and reviewed.

Digital mammography and DBT

Mammography is the first line of screening and reduces deaths from breast cancer by 60% in women who choose to participate in screening.¹⁰ However, cancer detection on mammography relies on changes in architecture or density that may be subtle or absent in cases of ILC. The most common mammographic appearance of ILC is a mass with spiculated or indistinct margins (44% to 65% of presentations).¹¹ Invasive lobular carcinoma may also present as architectural distortion without a mass, a one-view asymmetry, a focal asymmetry, or shrinking breast over time (Figure 1).¹¹ Less commonly, ILC may present as microcalcifications, a round circumscribed mass, or with nipple retraction. In 1 series, albeit with a small number of patients with ILC, 85% of lesions were reported to be of the same opacity as normal fibroglandular tissue.¹²

A retrospective study of ILCs detected by screen-film mammography found that 31 of 40 (77.5%) were, in retrospect, visible on prior mammograms, which is evidence of their subtle appearance.¹³ A case-control series by Porter et al found that ILC was more likely to present as an interval cancer than a screen-detected cancer, with an odds ratio of 1.9 compared with other histologic subtypes.¹⁴ Additionally, the larger tumor size of ILCs identified at screening compared with other histologic subtypes suggests that these may grow for a longer time and remain undetected by imaging for a longer period.⁴

Retrospective series show high false negative rates for mammography in ILC. A series of 341 pure ILCs, of which 98% were palpable, yielded a 43% false negative rate of screen-film mammography.¹⁵ A study of 361 women with ILC, of which 256 ILCs were symptomatic and 105 were screen-detected, showed no significant difference in the false negative rate for screen-film mammography (28% false negatives) and digital mammography (DM; 34% false negatives).¹⁶

Large screening trials comparing DBT with DM for detection of ILC have shown that DBT has increased specificity and sensitivity.^17-19 The use of DBT also improves conspicuity of ILC compared with DM.^20-22 A retrospective review of 17 patients with ILC found that 53% could be seen on DM and DBT, 18% could be seen only on DBT, and 29% were occult on DM and DBT.²³ A prospective trial using independent double reading found that significantly more ILCs could be detected on DBT than DM (28/30 reads vs 19/30 reads).²⁴ This trial also found that ILC was identified by both readers in only 10/30 of DM studies vs 16/30 DBT studies.²⁴

The sensitivity of mammography is lower for ILC than for other histologic subtypes. The STORM trial was a large prospective screening study that showed overall sensitivity of DBT was 85.5% for all histologies but was 80.0% for ILC.^18,25 The OVVV trial was a large prospective screening study that showed overall sensitivity of DBT was 82.7% for all histologies and 81.6% for ILC.¹⁹ Note that neither of these differences is statistically significant because neither study was powered to detect differences in sensitivity between histologies.

It is known that the sensitivity of mammography decreases with increasing breast density, with sensitivity approximately 98% in fatty breasts and 30% to 48% in extremely dense breasts.²⁶ A large study of data from >2 million screening examinations from 5 Breast Cancer Surveillance Consortium registries found that the sensitivity of DM for ILC decreases with breast density, with sensitivity of 88.2% in nondense breasts and 73.4% in dense breasts.²⁷ The researchers found that DBT improves ILC detection by 53% compared with DM in women with dense breasts, although sensitivity remains lower in dense breasts (84.3%) than in nondense breasts (90.8%).²⁷ Overall, sensitivity for ILC was 80.8% using DM and 87.9% using DBT. A survey of members of the Society of Breast Imaging revealed that radiologists are aware of the decreased sensitivity in dense breasts.²⁸ Radiologists perceive ILC as difficult to diagnose on mammography, with only 66.8% feeling confident diagnosing ILC on screening mammography in nondense breasts and only 25% feeling confident in dense breasts.

Modern screening trials of mammography have been structured as comparisons between DM and DBT. A summary of those that include details of cancer histology including ILC is represented in Table 1. Sensitivity of DM for ILC ranges from 54% to 80.8%. Sensitivity of DBT for ILC ranges from 75% to 92%.

US

Screening US is not limited by dense breast tissue and has shown incremental cancer detection rates of 2.0 to 2.7 cancers per 1000 screening US examinations after negative mammography.⁴⁰ The majority of supplemental cancers detected by screening US are invasive, early stage, node-negative cancers.⁴⁰ Supplemental screening with US decreases interval cancer rates, particularly in women with dense breasts.^40,41

The most common appearance of ILC on US is a hypoechoic irregular mass with posterior shadowing (Figure 2).¹¹ Invasive lobular carcinoma may also have a subtle appearance on US, including posterior shadowing without a distinct mass, and up to 39% present as nonmass lesions.^11,42,43 It has also been reported that the size of ILC is underestimated by US.⁴⁴ Despite this, early retrospective case series of ILC showed 88% to 98% sensitivity of US.^16,45,46

Modern large screening studies vary in their reported sensitivity for ILC, depending on methodology and whether sensitivity of US is compared with mammography, MRI, or interval cancers. Also, note that studies of screening US are limited by low numbers of ILC observations. The largest retrospective study of 21 200 screening rounds of DBT and hand-held US (HHUS) reported that ILCs represented 13% of diagnosed cancers, and all were seen by HHUS (sensitivity of DBT alone not reported).⁴⁷ In this study, interval cancers were not reported, and sensitivity of US was only compared with DBT.

ACRIN 6666 was a multi-institution study of 2809 women at elevated risk for breast cancer and with dense breasts, who underwent 3 rounds of DM and HHUS screening. A subset of 612 women underwent MRI screening at the end of the 3 rounds and had complete data for analysis. Twelve participants were diagnosed with pure ILC during the study period. Of those, 1 was detected only by DM, 5 by DM and HHUS, and 5 by HHUS only; 1 was not detected on study imaging, and none were detected by MRI alone.⁴⁸ Sensitivity of DM for ILC was therefore 50%, and sensitivity of HHUS was 83.3%.

The DBTUST study was a multicenter study of 6179 women who underwent 3 rounds of annual screening DBT and HHUS screening.⁴⁹ Over the 3 rounds, 19 total ILCs were diagnosed: 6 by DBT only, 2 by DBT and HHUS, 5 by HHUS only, and 1 by a second reader on HHUS only (false negative by the first reader). There were 5 interval ILCs: 1 diagnosed by MRI and 4 presenting clinically. The 4 interval ILCs presenting with breast symptoms were reported as measuring 1.6 cm, 3.0 cm, 6.8 cm, and 11.0 cm. The researchers conclude that ILC remains a challenge to detect even with combined DBT and US. When considering interval cancers, sensitivity of DBT and HHUS were each 33%.

The Adjunct Screening with Tomosynthesis or Ultrasound in Women with Mammography-Negative Dense Breasts (ASTOUND) and ASTOUND-2 trials specifically evaluated the usefulness of screening US after DM and after DBT.^50,51 Between these 2 studies combined, there were 8 ILCs identified by HHUS after negative DM, only 2 of which were also identified by DBT. Therefore, screening US increases ILC detection even after DBT.

Automated whole-breast US (ABUS) and HHUS have similar rates of supplemental cancer detection and ILC detection. In small comparison studies, HHUS missed 1 ILC that was identified by ABUS, while another study found 2 ILCs on HHUS that were missed on ABUS.^52,53

Interval cancers with ILC histology decrease after supplemental US screening. In an observational cohort study of 6081 screening mammograms with same-day screening US that were propensity-score matched to 30 062 screening mammograms without same-day screening US, Lee et al found no significant decrease in interval cancer rates between both groups.⁵⁴ However, they reported that the percentage of interval cancers was 14.2% ILC in the mammography-alone group (20 ILCs out of 141 invasive cancers) vs 5.0% in the supplemental US screening group (1 ILC out of 20 invasive cancers). Similarly, in a trial of 66 680 women who underwent both DM and HHUS screening, ILC accounted for 7.1% of interval cancers diagnosed within a 12-month follow-up period (2 ILCs out of 28 cancers).⁵⁵

Table 2 summarizes supplemental screening US studies that report number of ILCs separate from other invasive cancers. The screening US studies reported here show that ILC represents 16% of additional cancers detected after negative DM and 25% of the additional cancers detected after negative DBT. The incremental ILC detection rate for US is 0.4 to 0.6 per 1000 examinations.

MRI

The infiltrative growth pattern of ILC is more effectively visualized on breast MRI compared with mammography and US. Similar to IDC, ILC most often appears as an irregular, spiculated mass (Figure 3).^6,58 Additionally, it may present as a nonmass enhancement, with studies showing variable percentages of nonmass presentation (5% to 69%).⁶ When evaluating kinetics, ILC is more likely to exhibit delayed enhancement rather than initial enhancement, although peak enhancement levels are comparable with other types of breast cancer.⁵⁹ Similarly, in semiquantitative analysis, ILC shows lower transfer constant (k^trans) values than IDC, indicating relatively reduced vascular permeability.⁵⁸ Overall, these findings suggest that ILC may be potentially better visualized on delayed postcontrast sequences (Figure 3E). However, MRI remains highly sensitive for detecting ILC, with one meta-analysis reporting an overall sensitivity of 93.3% for detecting ILC on MRI, similar to that for all cancer types detected by breast MRI. The authors suggested that, excluding 1 older study that used a section thickness of 4.2 mm, the sensitivity could potentially increase to as high as 96%.

Most studies of MRI screening for ILC are retrospective reviews of biopsy-proven lesions. Prospective data, mostly from recent trials of abbreviated breast MRI (AB-MRI) protocols, show a varying proportion of ILCs. Table 3 summarizes MRI screening studies for ILC. Kuhl et al evaluated 443 patients by comparing a full breast MRI protocol with an AB-MRI contrast-enhanced protocol and identified 3 ILCs among 11 cancers, all of which were visible on both protocols.⁶⁰ The authors note that when readers interpreted only the maximum intensity projection, the only missed cancer was a 4-mm ILC. In a subsequent study focusing on average-risk women across 3 screening rounds, the same group found 11 ILCs among 61 total cancers, with 10 detected in the prevalence round and only 1 in the incidence round.⁶¹ Similar findings were reported in a recent update to the DENSE trial. Bakker et al, using an abbreviated protocol in women with extremely dense breasts, identified 9 ILCs in 79 cancers, with 2 of 4 interval cancers being ILCs.⁶² These studies collectively indicate that supplemental breast MRI enhances the detection of ILC across varying breast densities. They also demonstrate that supplemental screening MRI detects additional ILCs in women at increased risk, in whom screening MRI is currently recommended, as well as in women at average risk.⁶³

Kwon et al observed that 3 of 9 missed cancers over 3 years of AB-MRI screening in 1975 women were ILCs, all smaller than 1.4 cm. Two were initially mischaracterized as background parenchymal enhancement, and 1 was not visualized on MRI even on retrospective review.⁶⁴ This suggests that the slower enhancement pattern of ILC may reduce its recognition during prevalence rounds of AB-MRI; ongoing trials providing long-term data might offer further insights.

Contrast-enhanced mammography

Contrast-enhanced mammography provides better cancer detection rates when compared with DM, DBT, and US and slightly decreased cancer detection rates but increased specificity compared with breast MRI for supplemental screening.^66,67 Invasive lobular carcinoma findings on CEM are similar to those on DM and DBT but with the addition of enhancement information from recombined images (Figure 4). A direct comparison of CEM and breast MRI in staging 72 women with known cancer found that 1 of 2 CEM occult lesions was a 2-cm palpable ILC seen as multiple enhancing masses on MRI; no ILCs were missed on MRI.⁶⁸ Additionally, highlighting its lower vascular permeability, a retrospective review of ILC on CEM in 44 patients found that ILC was more likely to demonstrate weak or moderate enhancement than IDC due to slower contrast uptake on both MRI and CEM.⁶⁹

Prospective studies of CEM have shown good detection of ILC, although ILC is only represented in small numbers. Jochelson et al evaluated CEM and breast MRI screening in 318 women at high risk, and both modalities detected 3 invasive cancers, including a 0.6-cm ILC presenting as nonmass enhancement on both modalities.⁷⁰ Sung et al retrospectively evaluated 904 women who underwent CEM for increased risk of breast cancer and found a cancer detection rate of 15.5/1000, including 2 subcentimeter ILCs seen only on recombined images. Two interval cancers were found, and neither were ILCs.⁷¹ Sorin et al evaluated 611 women undergoing CEM for increased risk screening and dense breasts with a cancer detection rate of 13.1 per 1000. One ILC that was visible both on recombined and low-energy images was found in the original 21 cancers.⁷²

The small retrospective staging studies suggest that sensitivity of CEM for ILC is slightly decreased compared with MRI. The 3 published prospective screening studies show high sensitivity of CEM compared with DM in high-risk women but are limited by low numbers of observations. Larger prospective studies comparing supplemental screening modalities in women at average, intermediate, and high risk are needed.

Future directions

Currently, there are no risk models to predict ILC, so women who would benefit from supplemental screening to identify ILC cannot be identified.²⁷ As personalized screening strategies are being investigated using imaging markers and other clinical markers of risk, researchers may consider evaluating risk for specific histologies or cancer types.⁷³

Artificial intelligence and radiomic tools may lead to improvements in ILC detection. Generally, deep learning tools have been shown to improve the accuracy of interpretation of screening mammograms without specific mention of histology.⁷⁴ An artificial intelligence decision-support tool retrospectively used on 83 ILC lesions showed 100% sensitivity, and researchers concluded that this tool may increase radiologist confidence when assessing small lesions on US.⁷⁵

The novel PET radiotracer F16a-[¹⁸F]fluoro-17b-estradiol (FES) was Food and Drug Administration approved in 2020 for detection of recurrent or metastatic ER-positive breast cancer. Because 95% of ILCs are ER-positive, FES-PET may be specifically useful in staging ILC. Additionally, a pilot study has shown that FES-PET/CT detects additional ILC lesions in the ipsilateral breast, contralateral breast, and axillae that were not identified on standard of care imaging in patients with known ILC.⁷⁶ Larger trials are needed to clarify the role of FES-PET/CT in the evaluation of patients with ILC.

Conclusion

Invasive lobular carcinoma is more difficult to detect than other histologic subtypes of breast cancer on all imaging modalities. Digital breast tomosynthesis improves ILC detection compared with DM, particularly for women with dense breasts. Screening US increases the cancer detection rate for ILC compared with DM or DBT alone. MRI is the most sensitive modality for the detection of ILC. The sensitivity of CEM screening for ILC has not been much evaluated in large screening trials, but early results demonstrate higher sensitivity than DM and DBT and lower sensitivity than MRI.

There remain gaps in the knowledge regarding performance of screening modalities for ILC. The sensitivity of screening modalities for ILC specifically requires screening trials to report histopathology of cancers by method of detection. Additionally, trials that do not report interval cancers do not allow for true sensitivity measurements but only reveal sensitivities of modalities relative to one another. It would also be instructive to report ILC rates over multiple screening cycles because ILC may grow undetected for a longer period than other cancers.

Invasive lobular carcinoma remains challenging to diagnose on conventional screening modalities, and future research initiatives, including increasing conspicuity on mammography and identifying patients who would benefit from supplemental screening, are required to improve early detection of ILC.

Funding

None declared.

Conflict of interest statement

None declared.

Author contributions

Beatriu Reig (Conceptualization, Data curation, Investigation, Methodology, Project administration, Visualization, Writing - original draft), and Laura Heacock (Data curation, Investigation, Writing - original draft)

References