Imaging diagnosis of metastatic breast cancer

Using a software based on statistical text mining and machine learning methods [3], we identified 33 relevant articles in PubMed on staging of advanced breast cancer, metastatic breast cancer, bone disease and distant metastases between 2009 and 2019. We identified papers in English language, both original studies and review articles. Additionally, we searched the references listed for additional relevant papers, using a total of 40 scientific papers.

The most common staging system for breast cancer is the American Joint Committee on Cancer (AJCC) TNM [4], which is based on tumour size and the degree of locoregional invasion by the primary tumour (T), the extent of regional lymph node involvement (N) and presence (or absence) of distant metastases (M) [5] (Table 1). M1 indicates the presence of any metastases to distant organs, implying a stage IV disease (regardless of the T or N status). Breast cancer may be stage IV at first diagnosis, or it can be recurrent from previous breast cancer. Stage IV disease showed a 5-year survival rate of approximately 22%, although this rate varies according to other factors, such as the hormone receptor status [6]. The median survival for patients with breast cancer and bone metastases is 65 months in the oestrogen/progesterone-receptor-positive (ER/PR-positive) groups, and 40 months in both the human epidermal growth factor receptor 2 (HER-2) positive and ‘triple-negative’ group [7].

Multi-modality imaging is widely used clinically for disease staging. However, all cancers are potentially systemic diseases and whole-body imaging techniques, such as whole-body hybrid imaging (PET-CT and/or PET-MRI) or whole-body magnetic resonance imaging (WBMRI) are increasingly performed to reflect this.

There are now effective lines of treatment for patients with metastatic disease, which can improve symptoms, prevent complications and prolong life [8‐13]. Furthermore, oligometastatic disease (typically < 5 metastases), may be suitable for aggressive local therapy in combination with systemic treatment. Hence, the combined assessment of local disease and whole-body staging, together with better understanding of the tumour molecular characteristics, is key to individualised treatment [14, 15].

The detection of breast cancer metastases in breast cancer varies according to disease stage. In early breast cancer, routine staging evaluations are directed at locoregional disease [16] as in stages T1 and T2 primary breast cancers, the incidence of distant metastases is < 2% [17] compared with 15–20% in stage T3 or T4 [18]. Accordingly, the American Society of Clinical Oncology (ASCO), the European Society for Medical Oncology (ESMO) and the Royal College of Radiologists (UK), in their clinical practice guidelines for breast cancer (updated in 2018, 2019 and 2014, respectively), do not recommend routine imaging for the M-staging of asymptomatic patients with early stage disease. Staging imaging studies are usually performed for patients at high risk of disease spread, such as stage T3/4 cancers (> 5 cm), in patients with 4 or more involved axillary lymph nodes or in the setting of recurrent disease [16, 19].

The incidence of bone metastases in ER/PR positive cancer, which may establish upwards of 10–20 years after initial diagnosis [20], is significantly higher than the incidence of other site of metastases, namely is reportedly 18.7% (in luminal A subtype) to 30.4% (in luminal B subtype) [21]. Furthermore, the molecular subtype of the breast cancer influences the likelihood of metastatic spread: women with ER/PR-negative tumours have a higher risk of metastatic relapse in the first 5 years [20, 21] compared with ER/PR-positive tumours.

Current international guidelines lack consensus as to whom and how to image for metastatic disease [22]. Traditionally, high-risk patients were screened for occult metastases using bone scintigraphy (BS), chest radiography and abdominal ultrasound or bone scintigraphy and CT of the chest abdomen and pelvis. However, the use of next generation imaging, such as hybrid imaging (PET-CT and/or PET-MRI) and WBMRI, has increased over the years [17]. Indeed, traditional conventional imaging frequently detects bone disease and visceral metastases in late stages, which are associated with poorer outcomes. Moreover, these methods often fail to demonstrate the heterogeneity of the tumour biology, leading to delay in the detection of treatment resistance and the opportunity for therapeutic modifications [15].

The ESMO’s guidelines recommend performing chest, and abdominal imaging (US, CT or MRI scan) and a bone scan can be considered for patients with clinically positive axillary nodes, large tumours (T3/4) or tumours with aggressive biology. If such methods are inconclusive, dual imaging methods combining functional and anatomical information such as ¹⁸F-FDG PET-CT are suggested [16].

The Royal College of Radiologists (UK) recommends staging with CT of the chest abdomen and pelvis for patients with large (T4) tumours or with heavy lymph node burden (N2 disease) with or without bone scan and a PET-CT for suspected inflammatory breast cancer.

The latest North American National Comprehensive Cancer Network (NCCN) guidelines [23] recommend BS, abdominal CT/MRI (including the pelvis if symptomatic), chest CT/¹⁸F-NaF PET-CT, in symptomatic patients or in stage I-IIB breast cancer and abnormal liver function test, elevated serum alkaline phosphatase, localised bone pain. The ¹⁸F-FDG PET-CT is often recommended when the findings of conventional imaging are suspicious or uncertain. BS or ¹⁸F-NaF PET-CT may be bypassed when ¹⁸F-FDG PET-CT has already detected skeletal metastasis. Other requests for imaging may result from multi-disciplinary team discussions, e.g. in patients with triple negative invasive carcinoma, or in ipsilateral recurrence within the breast [24].

Regarding bone metastases, every above-mentioned imaging modality evaluates different aspects of the tumour: BS estimates osseous remodelling and osteoblastic activity, CT reveals bone destruction and/or presence of sclerosis, diffusion-weighted MRI assesses tissue cellularity and PET-CT using FDG tracer evaluates increased glycolytic metabolism [25]. Conventional imaging is limited when detecting small bone metastases: BS may have an unsatisfactory performance for lytic lesions, metastases with low bone turnover and low vascularity. CT usually demonstrates lytic lesions associated with bone destruction, but disease confined to the bone marrow may be missed [25].

In the next section, we review the diagnostic utility of conventional imaging (BS, CT and MRI) and next generation imaging (PET-CT, PET-MRI and WBMRI) for assessing the presence, extent and biological characteristics of bone and visceral metastases in patients with breast cancer.

In metastatic breast cancer, malignant cells displace and replace normal bone marrow fat cells causing a reduction in fat content, whereas successful management is associated with return of healthy bone marrow fat [58]. MRI can differentiate healthy from pathological bone marrow acting as an effective and non-invasive test to recognise bone metastases [17, 28, 59, 60]. MRI sequences including T1WI, diffusion-weighted imaging (DWI) and Dixon quantitative chemical shift imaging (which estimates water and fat fraction) can evaluate the anatomical and functional features of bone marrow [58, 61]. Therefore, a trained radiologist can discriminate bone metastases (which manifest with nodular focal metastatic lesions or marrow infiltration/replacement) from benign marrow alterations (such as marrow hyperplasia induced by chemotherapy) [28].

Studies evaluated MRI as the primary screening technique for diagnosing skeletal deposits to associate the advantages of high sensitivity and specificity with improved spatial resolution, demonstrating the diagnostic superiority of MRI over BS for identifying bone metastases [62, 63]. Particularly, an updated meta-analysis showed a pooled sensitivity of 97% for MRI versus 79% for BS and a pooled specificity of 95% versus 82%, respectively [29].

Additionally, MRI is commonly used to clarify equivocal findings from BS [28] and to detect complications associated with bone metastases (such as compression of spinal cord or nerves), which can alter management decisions [64]. Altehoefer et al. [65] showed how MRI guided the need for local therapy in sites that were negative on BS; Kim et al. [66] highlighted the value of MRI for the evaluation of single ‘hot spots’ on BS [67]. However, interpretation of bone MRI needs an understanding of the evolution of normal bone marrow with age: indeed in younger patients, highly cellular hematopoietic marrow may make it more difficult to diagnose small metastases [68].

With an 80–82% and 96–98% reported ranges of sensitivity and specificity, respectively [49‐51], MRI is an appropriate modality to detect hepatic metastases [69]. Breast cancer metastases in liver are typically hypo- to isointense on T1WI, iso- to hyperintense on T2WI and, since breast metastases are frequently hypovascular, they usually show perilesional enhancement [70] in the arterial phase of contrast enhancement. However, hypervascular metastases may also occur, which can be associated with disease progression (Fig. 4). In addition, DWI can aid the detection of small liver metastases easily overlooked on other sequences [69]. Pretreatment enhancement features of liver metastases at MRI, such as the degree of hypervascularity of the tumour rim, may also predict disease progression [71].

For the detection of brain metastases, contrast-enhanced MRI has higher sensitivity than CT [72], due to its higher soft tissue resolution and its better detection of parenchymal and leptomeningeal involvement [21]. Lesions are usually supratentorial, and they can be solitary or multiple, arising at the grey-white matter junction and at watershed zones of major arterial territories [73] (Fig. 5). Moreover, MRI showed higher sensitivity and specificity than PET-CT for brain metastases [74] as the high background activity present in the cortex and basal ganglia (due to intrinsic high glucose consumption of these structures) can substantially degrade signal-to-noise ratio of FDG PET [72], therefore adversely affect the ability of PET to detect especially small metastatic lesions.

Nevertheless, according to the updated ASCO’s guidelines, clinicians should not perform routine MRI to screen all the breast cancer patients for brain metastases but only in patients with HER2-positive advanced breast cancer because their high incidence of brain metastases [75].

Lungs are frequently involved by breast cancer metastases: with advanced MR techniques, lung metastases may be visualised as foci of high signal on DWI [10].

Among the various imaging modalities currently available for distant metastases detection, hybrid techniques which fuse morphological and functional data are the most sensitive and specific, and PET-CT and PET-MRI will probably continue to evolve and become increasingly important in this field. Different imaging modalities are often used in combination to optimally detect bone metastases. Although conventional imaging has frequently been used for cancer staging, these methods have been found to be less sensitive and less specific in providing accurate data in a clinically relevant time frame. Owing to positive recent developments in advanced imaging, the current trend is toward whole-body imaging in a single session. The choice of modality is usually based on the clinical situation and the type of primary tumour. Further research is warranted to further address the impact of these costly and labour-intensive imaging methods on treatment strategies and on the course of illness. Below, we discuss recent evidences on advanced imaging methods including PET-CT, PET-MRI and WBMRI. These modalities generally show improvements in diagnostic accuracy for detection of metastases over conventional imaging methods, with the ability to quantify biological processes related to the bone microenvironment as well as tumour cellularity. Comparative studies between conventional and advanced imaging techniques have also been carried out in subjects selected for further investigation after bone scan, and the contribution of hybrid imaging and WBMRI to evaluate bone/visceral lesions or suspicious symptoms have been reported [67].

The use of hybrid imaging (PET-CT and PET-MRI) has increased thanks to the increased access to commercial radiopharmaceutical production facilities and to the widespread availability of scanners [76]. The net clearance of 18F-sodium fluoride (¹⁸F-NaF) in breast cancer bone metastases is 3–10 times greater than that in healthy bone, conferring the capacity to detect both osteolytic and osteosclerotic metastases [77]. The mechanism of uptake of 18F-NaF tracer into bones is like that of ^99mTc diphosphonates, being related to local blood flow and osteoblastic activity, with rapid initial uptake into bone mineral as fluorapatite. ¹⁸F-NaF PET-CT has better diagnostic performance than ¹⁸F-NaF PET alone without the CT component [78].

In oncological staging, 18F-fludeoxyglucose (¹⁸F-FDG) is the most widely used radiotracer [76]: its uptake in skeletal metastases is pretended to be largely within tumour cells acting as a breast cancer-specific tracer rather than reflecting alterations in the bone microenvironment [79] (Fig. 6).

The Royal College of Radiologists (UK) recommends the consideration of PET-CT for staging breast cancer patients in those with multifocal disease, suspected recurrence when breast MRI is not possible and in symptomatic patients with an equivocal MRI. In addition, breast cancer patients under the age of 40 may benefit from PET-CT in early disease as this may result in upstaging of systemic sequelae [19].

The underlying histologic subtype of breast cancer may also influence the choice of PET imaging. Untreated invasive lobular carcinoma may have poorer ¹⁸F-FDG uptake of the osteoblastic metastases compared with invasive ductal or mixed subtypes [80]. Similarly, previous treatment history is significant, as ¹⁸F-FDG–negative skeletal metastases may show increased sclerotic following successful systemic therapy, which renders tumour cells nonviable even though ongoing reparative osteoblastic activity, as detected by BS or ¹⁸F-NaF PET may endure [81].

Nowadays, the processes of breast cancer metastasis have been well characterised at the molecular level and numerous biomarkers of tumour aggressiveness have been discovered and are ready to be tested in clinics. As we have discussed, molecular imaging offers the opportunity to depict specific cell markers relevant to tumour aggressiveness. For instance, Harmon et al. [111] recently compared in a prospective way the ¹⁸F-NaF PET-CT with ¹⁸F-N-[N-[(S)-1,3-dicarboxypropyl]carbamoyl]-4-F-fluorobenzyl-L-cysteine (DCFBC) PET-CT in patients with prostate cancer. Interestingly, this study appears to show that such tracer uptake depends on the disease course and treatment status, which may indicate a functional difference in bone metastases. The study showed that ¹⁸F-DCFBC PET-CT detected significantly less bone lesions, especially in patients who were found to be castrate sensitive. On the other hand, in more advanced disease, there was good concordance between the two radiotracers. ¹⁸F-DCFBC, though, demonstrates a high blood-pool uptake and such second-generation tracer may elucidate more functional information on the pathophysiology of bone metastases.

Although target-specific molecular imaging probes for tumour invasiveness have been developed for PET (i.e. proteases associated with tumour invasion, such as specific matrix metalloproteinases or cathepsins, can be targeted ‘in vivo’ with PET), they have not yet been widely used with MRI [112]. Novel MRI contrast agents based on iron oxide and dendrimer nanomaterials allow for better characterisation of tumour metastases [112]. Particularly, ultrasmall superparamagnetic particles of iron oxide (USPIOs) imaged with MRI does not require ionising radiation, yet can detect small metastases [113]. The rationale for the use of USPIO is that, after intravenous injection, the nanoparticles are phagocytosed by macrophages in circulation which then enter the interstitial space and are taken up by lymphatics [114‐117]. Since iron oxide is superparamagnetic, it becomes strongly magnetic in the strong magnetic field of the MRI leading to spin dephasing and susceptibility effects which result in signal loss. Therefore, the signal intensity is markedly reduced in healthy tissues due to the magnetic susceptibility and T2WI shortening effects of the USPIO particles [114, 118]. Conversely, in areas of metastases, there is much less uptake of USPIO particles and, therefore, those portions of the LNs remain unchanged in signal on T2WI 24–48 h after intravenous injection of a USPIO [114‐117]. Current limits of USPIOs’ use include the difficulty of image acquisition/interpretation and the lack of approved USPIOs themselves which are clinically available hinders adoption and larger studies.

Organ-specific MRI contrast agents are also used to identify metastatic disease in the liver. Superparamagnetic iron oxide particles (SPIO) have been approved for clinical practice use as a ‘negative contrast agent’ for the normal liver parenchyma in order to visualise benign and malignant hepatic cancers [119]. Dendrimer-based macromolecular MRI contrast agents as diaminobutane (DAB) dendrimers, that differ from polyamidoamine (PAMAM) dendrimers because they have a pure aliphatic polyamine interior, homogeneously enhance the liver parenchyma and are excreted more rapidly through both the liver and kidney than the analogous PAMAM dendrimer of similar molecular size [120]. Kobayashi et al. were able to detect small metastatic lesions on the liver from colon cancer cells using DAB dendrimers [120]. Additionally, gadolinium ethoxybenzyl diethylenetriaminepentaacetic acid (Gd-EOB-DTPA) has been developed as a liver-specific contrast agent by a small modification of conventional gadolinium-based chelates. As Gd-EOB-DTPA is taken up by hepatocytes, the normal liver parenchyma can be visualised by increased signal intensity at T1WI, generating a positive contrast enhancement, where normal liver parenchyma is bright on T1WI and the liver metastases are shown as focal areas of relative hypointensity [121].

Finally, hybrid imaging techniques of combining WBMRI with ¹⁸F-choline PET have recently showed to have a significantly higher sensitivity (93.5%) when compared to BS (63.6%) and WBMRI on its own (72.7%) [122]. The results are promising, but with limited access to PET-MRI and its high cost, a defined role of PET and newer advanced hybrid imaging techniques are needed [113]. Stratifying patients according to disease course and treatment status may prove beneficial in the future. Nevertheless, a hybrid role of functional imaging provided by biomarkers with the anatomical detail provided by imaging techniques will offer a valuable insight into disease status.