Numerous platforms were used to analyse and visualize WOSCC, such as the online analysis platform of bibliometrics (https://bibliometric.com/app, https://webofscience.com/wos), CiteSpace, and VOSviewer (Chen, 2004; Sood and Rawat, 2021). These were used to visualize and extract information from the collected database.

CiteSpace, a Java-based bibliometric application, was used to analyse dynamics and research clusters in preferred science topics (Chen, 2004) as it is compatible with the WOS data format. CiteSpace (version 6.2. R2) was used to visualize the distribution of countries/regions, contributions of institutions, authors, co-cited references, and co-occurring subject categories, and create a dual-map overlay of journals, the timeline view map, keyword analysis, and keyword and reference bursts.

VOSviewer is software that creates bibliometric maps (van Eck and Waltman, 2010). Three different visualization maps with different meanings were generated by VOSviewer, including network, density, and overlay visualization. VOSviewer (version 1.6.17) was used to visualize country/regional cooperation, author co-authorship, and keyword co-occurrence. A node on maps represents a specific parameter. Its size is decided by weighing the attributes. The clusters, links, and level of total link strength (TLS) reflect the strength of connections; TLS represents the level of strength between the connection nodes, which is weighted by the thickness of the lines.

We collected studies, annual publications, countries/regions, institutions, authors, articles, journals, references, and keywords. Retrieved articles and reviews were saved as plain text files and exported to EndNote Desktop as full-text citation records, ultimately called “download_XXX.Txt.”

Great progress has been made over the last decade in our knowledge of TME. In comparison with normal tissue, the TME of solid tumours possesses several other particularities, such as hypoxia, acidic pH, reactive oxygen species (ROS), elevated ATP, excessive Zn²⁺, a high level of glutathione (GSH), and higher levels of certain enzymes (Du et al., 2015; Pei et al., 2022). Owing to the increased rate of lactic acid production, the TME of solid tumours is more acidic than that of normal tissue (Webb et al., 2011). Moreover, the aberrant vascular network of tumours disrupts sufficient blood supply to all cells in the tumour mass, leading to hypoxia (Brown and Wilson, 2004). Aerobic metabolism leads to the production of ROS, which are generated by the mitochondrial respiratory chain. ROS levels are substantially higher in diseased than in normal tissues (Pei et al., 2022). In addition, there is a significantly higher amount of GSH in malignant tumour cells, leading to frequencies 1,000 times higher than that in healthy cells (Yin et al., 2018). Furthermore, enzymes such as matrix metalloproteinases are upregulated in TME. These enzymes not only promote cancer cell invasion and metastasis by clearing the pathways of invading cancer cells but also provide nutrients for vascular tumours (Overall and Kleifeld, 2006; Radisky et al., 2017). In recent decades, substantial advances in the development of TME-responsive therapy have offered promising diagnostic and therapeutic strategies. They can exhibit dramatic changes in developing TME-responsive nanomaterials such as pH, temperature, light, reduction/oxidation, and certain enzymes (Uthaman et al., 2018; Wang et al., 2021b).

The actual anti-tumour effects of nanotherapeutics are limited by various TME factors, such as tumour hypoxia, heterogeneity, and endosomal escape (Uthaman et al., 2018; Hu et al., 2021). Cheng et al. (2022b) developed an NDDS to achieve TME-responsive and targeted delivery of DTA-encoded plasmids (pDTA) to tumour sites via dual targeting to clusters of differentiation-44 and α_vβ₃ receptors. They subsequently used a combination of losartan and the DTA expression light-switchable transgene system to treat BC based on the NDDS. In vivo studies have demonstrated a decrease in active transforming growth factor-β and collagen type I, deeper tumour penetration, and increased survival rate after the novel NDDS application. Abumanhal-Masarweh et al. (2019) reported that treating BC with liposomal bicarbonate combined with a sub-therapeutic dose of Dox achieved excellent therapeutic outcomes, compared with Dox or bicarbonate monotherapy in mouse models of BC. In addition, 100-nm liposomes loaded with sodium bicarbonate were used as adjuvants to elevate TME pH. Liu et al. (2022) administered silibinin and curcumin co-loaded nanoparticles into 4T1 tumour-bearing mice and received an excellent response by inhibiting BC metastasis. Zhang et al. (2021a) designed a novel synergistic cascade strategy (SCS) that involved the use of mild hyperthermia and a smart drug delivery system (SDDS) to alter TME resistance for the effective therapy of TNBC. Shen et al. (2018) engineered, formulated, and delivered genes encoding an IL-10 protein trap to change immunosuppressive TME. This protein trap effectively inhibited the metastasis of TNBC, as shown in Figure 9’s summary of TME of solid tumours and TME-related targets.

As well as their specific targeted effect, the representation of TME in tumour models can lay the foundation for directing combinatorial treatment strategies and cancer nanotherapeutics. Sethi et al. (2015) developed a novel 3D co-culture spheroid model (3D TNBC) representing the tumour milieu of TNBC in vitro. Ozcelikkale et al. (2017) developed the TME-on-chip and assessed the delivery and efficacy of Dox in small molecular form versus hyaluronic acid-based nanoparticle (NP) formulation in MCF-7 and MDA-MB-231. Two cell lines were representative of different molecular subtypes of BC.

At present, research on TME-related targets is expanding. Researchers have designed a series of NDDSs based on TME factors and TME-related targets (e.g., tumour-associated macrophages (TAMs) and fibroblast activation protein on cancer-associated fibroblasts), which can be easily functionalised for specific targeting (Ji et al., 2016; Cheng et al., 2021). TAM-directed radiotracers and iron oxide NPs are promising tools for targeting TME and for monitoring the effects of immunotherapy via PET and MRI (Mukherjee et al., 2019).

The application of NDDSs to the treatment of cancer has recently gained interest (Liyanage et al., 2019). NPs can deliver drugs more selectively for the treatment of both primary and metastatic cancers. They can help overcome drug resistance, reduce side effects, enhance the bioavailability of poorly water-soluble drugs, and improve drug efficacy (Yap et al., 2021). NDDSs target cancer cells through “passive” and “active” mechanisms. In passive targeting, NPs without targeting ligands accumulate in the tumour interstitial space via the enhanced permeability and retention (EPR) effect (Attia et al., 2019). Active targeting is achieved by the binding of molecular ligands onto the surface of NPs to specific receptors on the tumour cell membrane, which are overexpressed (Patra et al., 2018). Various types of NDDSs with certain advantages have been used in the treatment of BC (Figure 10). The following sections classify these NDDSs based on the structure of NPs and the application of various materials in nanomedicine. The most prevalent classes of nanomedicines are listed in Supplementary Table S6.

As mentioned previously, BC is projected to be the most common cancer and the second leading cause of cancer-related death among women in 2023 (Siegel et al., 2023). BC is diagnosed based on standardized pathological criteria in clinical settings. Approximately 50%–75% of BC cases are attributed to invasive ductal carcinoma, whereas 5%–15% of cases are attributed to invasive lobular carcinoma (Dillon et al., 2014). The remaining cases are classified under special histological types, such as mixed ductal/lobular carcinomas (Dillon et al., 2014). BC comprises 20 morphologically distinct subclasses, such as cribriform, papillary, spindle cell, solid, neuroendocrine, clinging, medullary, clear cell, secretory, and mucinous (Hsiao et al., 2010). At present, molecular classification is widely accepted, whereas morphological classification is uncommon. Two main molecular targets have been identified: oestrogen receptor alpha (ERα)/steroid hormone progesterone receptor (PR) and epidermal growth factor 2 (ERBB2, HER2, or HER2/neu). ERα is a steroid hormone receptor and transcription factor. It expresses in approximately 70% of invasive BC, activating growth pathways. PR is also a principal marker of ERα signalling (Joshi, 2018). The expression of ER or PR is defined as HR+, when 1% of tumour cells are over-expressed. Additionally, ERBB2, HER2, or HER2/neu is a transmembrane receptor tyrosine kinase in the epidermal growth factor receptor family. It is significantly overexpressed in approximately 20% of BC. The amplification of the gene ERBB2 is defined as ERBB2+ (Wolff et al., 2007). TNBC is characterized by a lack of expression of ER, HER2, or PR and accounts for approximately 15% of BC cases (Denkert et al., 2017). These three molecular subtypes of BC have distinct prevalence rates and prognoses and require different systemic treatments. According to TNM classification, BC is divided into four stages. Stage I is the presence of breast tumours of <2 cm in size without lymph node involvement, whereas Stage IV is defined by metastasis from the breast and axilla to distant sites—most commonly bones, brain, liver, and lungs (Chavez-MacGregor et al., 2017) (Figure 11). Stage IV BC accounts for approximately 6% of all BC cases in the United States (Lee et al., 2020).

The progress of BC treatment is not limited to surgery, chemotherapy, immunotherapy, endocrine therapy, targeted therapy, and radiation therapy (Figure 12; Supplementary Table S7). Eradicating regional lymph nodes and breast tumours to prevent metastasis and recurrence is the primary principle of BC treatment. Local therapy includes surgical resection, axillary lymph node sampling, or removal, followed by postoperative radiation. BC therapy can be preoperative or postoperative. Different systemic therapies are required for treating different BC subtypes, such as endocrine therapy for HR+ BC, trastuzumab-based antibody therapy, and chemotherapy for all ERBB2+ tumours and chemotherapy alone for TNBC. Maintaining basic survival and improving patient quality of life are primary approaches to treating metastatic BC. Adjuvant approaches are common between metastatic and nonmetastatic BC. Surgery and radiation are typically used to alleviate symptoms in metastatic BC (Waks and Winer, 2019).

Chemotherapy essentially includes adjuvant and neoadjuvant therapy in the treatment of BC. In this section, we summarise the significance of randomized clinical trials that investigate the efficacy of adjuvant and neoadjuvant chemotherapy (NAC) for BC (Supplementary Figure S3). NAC is recommended in patients with locally advanced disease or aggressive tumour biology before surgical resection (Leon-Ferre et al., 2021). The advent of adjuvant chemotherapy in BC was by Bonadonna et al. (1976). The first chemotherapeutic regimen prescribed was a combination of cyclophosphamide, methotrexate, and 5-fluorouracil (CMF). CMF was used as an adjuvant treatment to radical mastectomy in patients with primary BC with positive axillary lymph nodes and prevented metastatic recurrence (5.3% of 207 patients treated with chemotherapy vs 24% of 179 patients without chemotherapy). A 1990 study reported that the therapeutic effects of four cycles of Adriamycin and cyclophosphamide (AC4) demonstrated a non-inferiority margin for 6 months of CMF in patients with tamoxifen-nonresponsive positive-node BC (Fisher et al., 1990). A 1999 trial explored the effectiveness of concurrent versus sequential regimens and demonstrated that treatment with AC4 followed by four cycles of docetaxel (ACT) improved the survival of patients with invasive adenocarcinoma (tumour stage T1, T2, or T3; clinical nodal stage N0 or N1; and metastasis stage M0). Compared with doxorubicin–docetaxel and the concurrent-ACT group, the result of the trial of the sequential-ACT group was positive (8-year DFS: 74% vs 69% vs 69%, respectively). However, no significant improvement in OS was seen (8-year OS: 83% vs 79% vs 79%, respectively) (Swain et al., 2010). Until 2001, the role of trastuzumab in HER2+ BC was presented, especially for metastatic HER2+ BC. In the trial, trastuzumab plus anthracycline and cyclophosphamide (AC) vs AC-alone was used. The result demonstrated the therapeutic effects by progressive disease (7.4 months vs 4.6 months; p < 0.001), objective response (50% vs 32%, p < 0.001), duration of response (median, 9.1 vs 6.1 months; p < 0.001), and longer survival (median survival, 25.1 vs 20.3 months; p = 0.01) (Slamon et al., 2001). In a 7-year follow-up of US oncology research trial 9735, four cycles of docetaxel/cyclophosphamide (TC) were found to have better therapeutic effects than four cycles of an anthracycline (doxorubicin)/cyclophosphamide (AC) in early BC (DFS, 81% vs. 75%, respectively; HR, 0.74; OS, 87% vs. 82%, respectively; HR, 0.69) (Jones et al., 2009). In the 2017 ABC trials, the analysis of TC6 versus the TaxAC regimens (various anthracycline-plus-taxane-containing regimens) was planned, with invasive disease-free survival (IDFS) as the primary end-point for early BC. The TaxAC regimens were proven to not be non-inferior to various TC6 (4-year IDFS, 90.7% for TaxAC vs 88.2% for TC6; p = 0 .04) (Blum et al., 2017). In addition, the NSABP B-18 trial conducted was the first study to compare the outcomes of preoperative and postoperative AC in 1998. The results suggested that preoperative chemotherapy was as effective as postoperative chemotherapy, and that there was no significant difference in DFS, DDFS, and survival (p = 0.99, 0.70, and 0.83, respectively) between both groups. Meanwhile, many women have undergone lumpectomy (Fisher et al., 1998). NAC is the preferred therapeutic approach for TNBC and HER2-positive BC (Leon-Ferre et al., 2021). Chemotherapeutic drugs, including Dox and PTX, have also been used in combination with drugs such as disulfiram, thioridazine, and curcumin to improve therapeutic outcomes (Núñez et al., 2016).

Despite progress in the existing treatment strategies, the treatment of brain metastases has not advanced in parallel. The treatment of metastatic BC is challenging, partly attributed to the limited blood–brain barrier penetration of anti-neoplastic agents and genetic heterogeneity-activation of the PI3K/AKT/mTOR pathway involved (Dagogo-Jack et al., 2017; Bowman and Kumthekar, 2018). Toxicity associated with chemotherapeutic agents can directly or indirectly cause, for example, anaemia, cardiotoxicity, pulmonary toxicity, nephrotoxicity, and thrombocytopenia.

NDDSs may be modified to improve the treatment of BM in BC. However, efficient drug delivery to the tumour is still unsolved. Biological barriers to drug transport prevent the successful accumulation of chemotherapeutic agents at the lesion location. TME is a barrier to NDDSs since TME of solid tumours possesses particularities like hypoxia and acidic pH (Du et al., 2015; Pei et al., 2022). High interstitial fluid pressure, desmoplastic stroma, and TAMs hinder the diffusion of nanotherapeutics, thereby reducing the cellular uptake rate (Blanco et al., 2015; Cheng et al., 2021). The efficacy of chemotherapeutic drugs administered via NDDSs is severely limited owing to poor and nonselective cellular uptake and unstable circulation; further clinical trials are warranted (Wang et al., 2021c). In addition, NP transport is related to the administration route, which determines the biodistribution pattern of drugs. Oral, intradermal, and subcutaneous routes of administration should be further explored in future studies (Wang et al., 2021a).

The other major obstacle is the development of MDR. It can be categorized as primary (manifesting as tumour insensitivity to initial treatment) or acquired (occurring after initial response to therapy) (Perez, 2009). Chemotherapeutic agents have transient responses because of MDR (Gorain et al., 2018). Tumours may not respond to systemic therapies; in particular, metastatic BC does not respond to initial treatment. Consequently, many studies have used liposomes (Deng et al., 2014), NPs (Pan et al., 2013), micellar systems (Qiu et al., 2014), siRNA-targeted particular genes (Conde et al., 2013), and exosomes (McClements, 2020) to overcome MDR. NDDSs may offer additional benefits by overcoming the limitations of conventional treatment, including difficulties in targeting, dynamic in vivo changes in the materials, and multiple biological barriers. However, these experimental results need further testing.

As shown in Figure 13, enhancing drug delivery to tumours is urgently needed for many steps, including NP–protein interactions, retention of NPs in blood circulation, extravasation of NPs into the TME, penetration of tumours by NPs, cellular uptake and intracellular trafficking of NPs, and controlled drug release.

Adsorption of proteins on the surface of NPs in a biological milieu results in the formation of protein corona. Proteins on the surface of NPs continuously undergo the dynamic process of adsorption and desorption (Cai and Chen, 2019). These changes influence NPs’ surface properties, electric charges, hydrodynamic size, and their other physicochemical processes, including their corrosion and coagulation (Zarei and Aalaie, 2019). In addition, they affect cellular uptake, intracellular trafficking, PK, biodistribution, and toxicity (Shi et al., 2017).

Blood circulation refers to NPs that are distributed to a specific organ/tissue/cell through blood circulation. Therefore, blood circulation half-life—the efficiency with which NPs passively extravasate from the microvasculature into the TME—is an important pharmacokinetic (PK) parameter of NPs. NPs should be retained in blood circulation for a long period to ensure the delivery of adequate concentrations of drugs to target organs/tissues/cells; thus, long-circulating NPs coated with an inert polymeric material should be developed. PEGylated NPs offer a strategy for decreasing immunogenicity and prolonging circulation time, thus overcoming various biological barriers (Suk et al., 2016). PEG is usually incorporated onto the surface of liposomes or inserted after vesicle formation (Ho et al., 2013), limiting the direct interaction between various blood components and liposomes (Takakura and Takahashi, 2022). The other major factor limiting circulation time is the non-specific interaction between NPs and serum proteins (protein corona formation). The composition and quantity of protein corona largely depend on the physicochemical properties of NPs, especially their in vivo pharmacokinetic properties (Bertrand et al., 2017). Rodriguez et al. (2013) developed nano-formulations that used minimal “self” peptides attached to virus-size particles. Intravenous injection of this nano-formulation into mice resulted in persistent retention that enhanced dye and drug delivery to tumours.

NPs enter the tumour site from systemic circulation through extravasation, which can be influenced by the perivascular TME and aberrant tumour vasculature. Because of the metabolic demands of rapid cancer cell division, the tumour neovasculature is quite distinct from the normal vasculature and exhibits “leakiness.” Some vascular mediators such as nitric oxide and angiotensin II may enhance the extravasation of NPs from tumour blood vessels (Seki et al., 2009; Matsumoto et al., 2016). Therefore, normalising the tumour vasculature can improve anti-tumour efficacy and improve drug delivery efficiency (Magnussen and Mills, 2021). Iron oxide NPs, cuprous oxide, AuNPs, silica NPs, MAN-PLGA, and APRPG-PEG-ZOL-CLPs have been used to regulate the normalisation of tumour vasculature in BC models (Liang et al., 2022).

Tumour penetration can affect NPs in DDSs, hindering their penetration into the tumour core region and limiting their diffusion. The dense interstitial matrix (Li et al., 2018) and high abundance of perivascular stromal cells (Miller et al., 2015) in the TME impede NP penetration into the tumour site (Shi et al., 2017). Zhang et al. (2020) fabricated small morph NPs (PDMA) by modifying polyamidoamine (PAMAM) dendrimers with dimethylmaleic anhydride (DMA). PDMA facilitates deep penetration through active, energy-dependent caveolae-mediated transcytosis.

Cellular uptake and intracellular trafficking mainly refer to enhancing the retention of NPs at the cellular level through internalisation pathways. For effective intracellular delivery, targeting the nanocarrier to a specific organ/tissue/cell from blood circulation is the first step, after which NPs are internalised by cells via endocytic pathways. In addition, NPs can be optimised to enhance their cellular uptake and achieve specific targeting. NPs can also be optimised to enhance their cellular uptake and achieve specific targeting. For optimal intracellular delivery, NPs should be able to evade degradation, have effective and efficient targeting ability, and facilitate the controlled release of drugs into the intracellular environment. To this end, multiple actively targeted NPs have been developed (Prabaharan et al., 2009; Neves et al., 2013; Bilan et al., 2016; Park et al., 2017; Jin et al., 2018; Nowroozi et al., 2018; Patra et al., 2018; Sherje et al., 2018; Sun et al., 2018; Wang et al., 2018; Soltantabar et al., 2020; Al-Humaidi et al., 2021; Desai et al., 2021; Rommasi and Esfandiari, 2021; Giri et al., 2023). However, the cellular entry of nanocarriers is influenced by dynamic interactions between the carriers and the membrane. Therefore, carriers should be retrieved in the middle of the endocytic pathway, a phenomenon known as “endosomal escape” (Kim et al., 2019). Intracellular trafficking refers to the transport of cargo from the cell surface to its final destination within cellular compartments after cellular internalisation. NPs can be transported to various intracellular components. After cellular internalisation, NPs are confined within a membrane-lined vesicle, such as endosomes (Donahue et al., 2019). The partial cargo is transported to its destination by early endosome. Recycling endosomes with part of the cargo transports it to the plasma membrane. During maturation and differentiation, late endosomes come from early endosomes that integrate with lysosomes to form endolysosomal vesicles, and hydrolytic enzymes contained within these vesicles degrade the trapped NPs. Another intracellular degradation pathway that plays an important role in deciding the intracellular fate of NPs is autophagy. During autophagy, cytoplasmic contents are engulfed by autophagosomes and delivered to lysosomes to be degraded and recycled (Foroozandeh and Aziz, 2018). For accurate detection of endosomal escape, numerous nucleic acid- and peptide-based self-assembling nanomaterials, micelles, and peptide-conjugated QDs have been developed for intracellular delivery (Ruan et al., 2007; Kim et al., 2019; Yin et al., 2019).

Controlled drug release refers to the release of drugs during circulation. It contributes to the effective functional design and safe application of nanomaterials. Altogether, drug release rates, PK properties, and extravasation should be considered when designing NDDSs for achieving optimal outcomes. External stimulation may produce tailored drug-release profiles with temporal and spatial control (Meng et al., 2017; Yoon et al., 2020). Undifferentiated contact of NPs with physiological fluids results in the formation of protein corona that potentially shields surface functionalities (such as targeting ligands) and impedes drug release (Cai and Chen, 2019). In particular, targeted drug delivery plays an important role in stimulating responsive drug release (Chen et al., 2017). Given that the human body is a complex environment, further investigation is warranted to verify the effects of the in vivo release of drugs from NPs. Altogether, the therapeutic efficacy of NDDSs can be improved by selecting an optimal drug-to-carrier ratio and an adequate dosage of targeting ligands, thereby enhancing targeting ability and facilitating controlled drug release.

NDDSs can reduce systemic toxicity, realize efficient tumour targeting, have high bioavailability and drug solubility, and facilitate controlled drug release, thereby improving therapeutic outcomes (Din et al., 2017; Patra et al., 2018). To date, research in the field of nanomedicine has almost exclusively paid attention to tumour-targeted drug delivery. Nanocarriers that target tumours based on the EPR effect have been rapidly developed (Matsumura and Maeda, 1986). Many nanocarrier-based therapies have shown significant therapeutic efficacy in clinical settings (Khan et al., 2022; Park et al., 2022).

As shown in Figure 14, the data included papers published by PubMed and clinical trials registered at Clinicaltrials.gov. They were identified using the following keywords: “Polymer OR Liposome OR Metal OR Quantum Dots OR Gold OR Superparamagnetic Iron Oxide OR Silver OR Carbon OR Mesoporous Silica OR Protein OR Dendrimer OR Micelle OR Nanoparticle” AND “Breast Neoplasm” OR “Breast Cancer.” The specific keywords for nano-formulations were chosen as they represented the most nanomedicines assessed in clinical trials. Specific keywords for nano-formulations were selected as they represented the large majority of NDDSs assessed in clinical trials. A total of 465 clinical trials were identified (as of 1 July 2023). The distribution of clinical trials with respect to the clinical trial phases demonstrated that the majority are in Phase Ⅱ. To date, seven nano-formulations represented by Doxil^® have been approved for the clinical treatment of BC (Jiang et al., 2022).

The aforementioned findings suggest an increasing interest in developing next-generation nanocarriers with unique properties such as active tumour targeting and TME responsiveness. The use of active targeting for improving DDSs efficiency has revolutionised cancer therapy. In active targeting, NDDSs selectively interact with the specific overexpressed antigens or receptors on the tumour cell membrane, enhancing their cellular uptake (Ahuja et al., 2021; Feng et al., 2021). NDDSs can exhibit a dramatic change in specific properties in response to characteristics of TME such as pH, temperature, light, reduction/oxidation, glutathione levels, and an abundance of certain enzymes (Uthaman et al., 2018). To date, various NPs also have been used in the treatment of BC, including liposome NPs (Nowroozi et al., 2018; Abumanhal-Masarweh et al., 2019; Le Rhun et al., 2020; Liu et al., 2022), SLNs (Wang et al., 2018), PNPs (Soltantabar et al., 2020), metal-based NPs (gold NPs) (Jafarizad et al., 2017), SPIONs (Poller et al., 2017), QDs (Sun et al., 2018)), carbon-based NPs (Garriga et al., 2020), MSNs (Zhao et al., 2018), protein-based NPs (Esfandiari et al., 2016), dendrimers NPs (Jain et al., 2019), and biological components such as exosomes (Al-Humaidi et al., 2021). With respect to SDDS, nano-scale hybrid liposomes coated with magnetoliposomes have been used for intracellular magnetic targeting in BC (Plassat et al., 2011).

Furthermore, nanotechnology may improve the therapeutic efficacy of radiotherapy and guide surgical tumour resection (Giri et al., 2023). PEGylated doxorubicin liposomal formulation (Doxil^®/Caelyx^®) was approved by the FDA in 1995. This formulation can be used to treat BC. Moreover, much research in this field has already led to the completion of dozens of clinical trials. Nanomedicine can improve the quality of life of patients by reducing adverse effects (Melchor-Martínez et al., 2021). For example, Abraxane™ (protein-bound PTX, approved by the FDA in 2005) has demonstrated significantly higher efficacy than standard PTX in improving the outcomes of patients with metastatic BC. In addition, the efficacy of passively targeted nanocarriers is limited owing to the heterogeneity of the EPR effect and the physiological barriers associated with it. Recently, Lee et al. (2017) used ⁶⁴Cu-labelled HER2-targeted liposomes and PET/CT to assess the EPR effect with HER2+ metastatic BC in 19 patients.

Despite significant advances in nanotechnology, several challenges are associated with the development of NPs. The EPR effect and therapeutic outcomes of NDDSs are determined by multiple biological factors such as NP–protein interactions, blood circulation, extravasation into and interaction with TME, heterogeneity of EPR effect, TME-responsive delivery, and NP properties (e.g., size, geometry, and surface features). Furthermore, it is difficult to predict the safety and efficacy of NDDSs in humans based on preclinical animal models. In vitro and in vivo models of BC cannot accurately mimic the biological or clinical conditions of interest. On the other hand, selecting a target for actively targeted NCs is based on “classical” biomarkers (e.g., HER2 for BC). Actively targeted NCs can offer great advantages in treating BC. However, challenges presented by physiological barriers and MDR remain. Future studies should focus on developing optimal delivery systems for specific natural products (Yap et al., 2021). In addition, protein corona in the physiological environment, its consequences for biodistribution, and toxicity should be considered (Rosenblum et al., 2018). As shown in Figure 14, only a few nanomedicines reach clinical trials and market approval. More than 50,000 scientific research articles related to nanomedicine were published by the end of 2022; however, only seven nano-formulations have been approved for the clinical treatment of BC to date (Jiang et al., 2022). None of the active targeting nanocarriers have made advances in clinical trials (Rosenblum et al., 2018). The introduction of new DDSs is slow and expensive because clinical studies require time for rigorous assessment of safety and efficacy.

In this section, we summarise the seven core research themes. The application of nanotechnology in drug discovery and delivery offers novel strategies for overcoming the challenges associated with the traditional treatment of BC. We used CiteSpace to determine the future directions of research in the field of NDDS-based BC treatment. Keyword and reference bursts revealed highly cited articles related to NP-based treatment of metastatic BC (Cheng et al., 2022a). Keywords (“green synthesis,” “optimization,” “proliferation,” and “molecular mechanisms”) have recently received substantial attention. The five important articles related to these keywords were Hossen et al. (2019) (strength, 3.07), Rizvi and Saleh (2018) (strength 4), Patra et al. (2018) (strength 4.34), Golombek et al. (2018) (strength 4.93), and Rosenblum et al. (2018) (strength 6.17) (Figures 7D, 9E).

“Green synthesis” results in the production of NPs that are non-toxic, eco-friendly, economical, reproducible, and easily amplified and have well-defined morphology (Khalid et al., 2017; Fadeel et al., 2021). Therefore, NPs based on natural products and microorganisms have gained considerable attention in the field of NDDS-based BC therapy (Nasir et al., 2021). Rahimi et al. (2019) used a green synthesis method to prepare a novel amphoteric calix [4] arene (Calix) macrocycle coated on the surface of Fe₃O₄ magnetic nanoparticles and used as a magnetic nanocarrier for simultaneous delivery of Dox and methotrexate in MCF7 cells. “Optimization,” frequently mentioned as “formulation optimization,” refers to the modification of an old nanocarrier design to improve its targeting and drug delivery efficiencies. Oseni et al. (2021) encapsulated andrographolide in PLGA nanoparticles via emulsion solvent evaporation, which improved therapeutic efficacy in a metastatic BC cell line. In addition, NDDSs can be optimised using components such as nanovesicles (Kassem et al., 2018). Alhakamy et al. (2022) used these to assess the cellular uptake of quercetin and its inhibitory effects on MCF-7 cells. “Proliferation” refers to anticancer activities through the inhibition of cell proliferation against BC in in vitro research, providing a novel approach nanodrug carrier system (Yewale et al., 2018; Sun et al., 2021). Many researchers have investigated the molecular mechanisms underlying the anti-tumour effects of nano-formulations and NDDSs, especially for developing new and smart NDDSs (Hossen et al., 2019; Cheng et al., 2021; Yap et al., 2021).

As mentioned in “Nanotechnology as an emerging field in BC therapy: progress and challenges,” the clinical translation of NDDSs is hampered. According to reference bursts, overcoming the challenges associated with drug delivery is a novel research direction (Patra et al., 2018; Rizvi and Saleh, 2018). Researchers have addressed these challenges by improving the design of NDDSs. SDDSs that can overcome multiple drug delivery barriers represent a promising tool for improving therapeutic efficacy. The development of SDDSs is a research hotspot in the field of materials science and pharmacy. Smart drug carriers will have a huge impact on the clinical treatment of cancer in the near future (Hossen et al., 2019).

According to reference bursts, “progress and challenges toward targeted delivery of cancer therapeutics” is another hot issue in the frontier (Rosenblum et al., 2018). Selecting a target for actively targeted NCs is based on “classical” or disease biomarkers (e.g., HER2 for BC). Actively targeted NCs can offer great advantages in treating BC. However, challenges presented by physiological barriers and MDR remain to be overcome. Future studies should focus on developing optimal delivery systems for specific natural products (Yap et al., 2021). In addition, consideration should be given to the formation of protein corona in the physiological environment and to the efficacy, biodistribution, and toxicity of drugs (Rosenblum et al., 2018).

One research direction is assessing the heterogeneity of the EPR effect, first described in 1986 (Matsumura and Maeda, 1986; Golombek et al., 2018). The clinical outcome of nanomedicine is not as good as anticipated on the basis of preclinical results, suggesting that many nano-formulations fail to produce the desired effects (such as tumour accumulation) and question the existence of the EPR effect (Nichols and Bae, 2014). This effect has been demonstrated to be valid and reproducible in all tested animal models; however, animal tumour models do not accurately mimic the TME of human patients. Moreover, the poor quality and reliability of published preclinical studies impede the clinical development of nano-formulations (Duncan, 2017) with a failure rate of 90%–95%. The failure of Phase II–III clinical trials can be attributed to the lack of efficacy (52%) and safety (24%) as well as strategic (15%), commercial (6%), and operational (3%) reasons (Harrison, 2016). Possible solutions may include the selection of an appropriate pathological variety that suits specific nano-formulations, followed by the establishment of an ideal animal model that best mimics the human cancer type. Selecting the right set of patients and treating them with the correct nano-formulations are key steps to successful clinical translation. Therefore, when aiming to develop nano-formulations for clinical use, the heterogeneity of the EPR effect should be considered and strategies should be developed to overcome this obstacle. Future studies should focus on developing a combination of therapies that potentiate the EPR effect and diagnostic protocols that enable visualization and quantification of the extent of the EPR effect in individual patients (Zhao et al., 2017). Perhaps imaging EPR-based tumour targeting may provide such evidence and quantification of the EPR effect in metastatic BC (Zhang et al., 2017).

Although this bibliometric study visualized the depth of the systematic review, it has several limitations. First, we provided an overview of the evolution of NDDSs in BC treatment. We comprehensively reviewed ongoing developments and determined future research directions. This meant that we were limited to data held in specific databases, which were inevitably biased. Second, data were downloaded from the WOSCC database owing to the format constraints of the bibliometric analysis software. Therefore, articles present in other databases were missed, and those with significant impact may not have been included. In addition, this study included publications only from 2012 to 2022. Third, we extracted original research and review articles published in English and did not include articles published in other languages and adhering to non-research types, which may have led to some omissions. In addition, a uniform algorithm is not available for optimising parameters in CiteSpace and VOSviewer for bibliometric analysis. The outputs may vary slightly, with different settings leading to confusion among readers. Fourth, an accurate description of the diagnostic and therapeutic strategies of BC may apply to a certain group of people but differ internationally. Finally, the past 20 years have seen rapid development in NDDSs. Nonetheless, big data provides valuable insights and guides decision-making in any domain. Researchers can gain an in-depth understanding of current bottlenecks and the developmental status of the field of NDDSs via big data.