The Application of Magnetic Nanoparticles for Sentinel Lymph Node Detection in Clinically Node-Negative Breast Cancer Patients: A Systemic Review and Meta-Analysis

Simple Summary Generally, the standard method of applying sentinel lymph node biopsy in breast cancer patients is via a technetium-labeled nanocolloid (radiolabeled tracer) with or without blue dye. However, the radioactive agents may cause challenges both to hospitals and patients. Alternatively, a safer candidate method, such as SPIO, has been often introduced and validated in comparison with standard methods. The aim of this study was to perform a systematic review and meta-analysis to evaluate the diagnostic accuracy of SPIO and its clinical impact on the management of breast cancer. Based on our study, SPIO could be considered as an alternative standard of care for sentinel lymph node detection. Compared with the standard method, SPIO exhibited equivalent or even superior detection capacities, while safety could be guaranteed. Lower SPIO doses would also not impair detection capacity compared with the standard method. The absence of radioactivity of SPIO is one of the most important advantages for clinical applications. Abstract Superparamagnetic iron oxide (SPIO), an alternative mapping agent, can be used to identify sentinel lymph nodes in patients with clinically node-negative breast cancer. However, its performance in comparison with the standard method, using a radioisotope (technetium-99 m, Tc) alone or in combination with blue dye, remains controversial. Hence, a systematic review and meta-analysis were conducted to evaluate the diagnostic accuracy of SPIO and its clinical impact in the management of breast cancer. The PubMed, Embase, and Cochrane databases were comprehensively searched from inception to 1 May 2022. Cohort studies regarding the comparison of SPIO with standard methods for sentinel lymph node identification were included. A total of 19 prospective cohort studies, which collectively included 2298 clinically node-negative breast cancer patients undergoing sentinel lymph node identification through both the standard method and SPIO, were identified. The detection rate for sentinel lymph nodes (RR, 1.06; 95% CI, 1.05–1.08; p < 0.001) was considerably higher in the SPIO cohorts than in the standard method cohorts, although this difference was not significant in detected patients, patients with positive sentinel lymph nodes, or positive sentinel lymph nodes. Compared with the standard method, the SPIO method could be considered as an alternative standard of care for sentinel lymph node detection in patients with clinically node-negative breast cancer.


Introduction
Sentinel lymph node biopsy (SLNB) is routinely used to confirm the metastatic status of the axilla in patients who are clinically presented with a node-negative disease in most early-stage breast cancers. Generally, the standard method of applying SLNB is

Materials and Methods
We conducted a meta-analysis of cohort studies to assess whether SPIO is a potential substitute for the standard method for sentinel lymph node (SLN) detection in patients with clinically node-negative breast cancer. This study followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guideline [17]; the protocol was registered on PROSPERO (CRD: 42020218583).
The PubMed, Embase, Web of Science, and Cochrane databases were searched from inception to 1 May 2022. A combination of Medical Subject Headings and free-text terms, including their variants, was searched in those databases. The Medical Subject Heading (MeSH) terms used were (breast cancer' AND 'sentinel lymph node' AND ('nanoparticle*' OR 'magnet*' OR 'magnetic*' OR 'sentimag')). Search strategies were adapted to each database's search engine and were limited to those in the English language and involving human participants. All articles were reviewed based on their titles and abstracts, and those meeting the inclusion criteria were then read in full text. Complementally, the references of the included articles were also searched manually for new candidate articles. The detailed search strategies are shown in the Supplements (File S1).
Studies were included if they fulfilled the following inclusion criteria: prospectively conducted clinical trials in which the efficacies between magnetic technique and standard technique (using either radiolabeled tracer alone or a combination of BD) for SLNB were compared in patients with breast cancer; ethical approval obtained; sample size greater than 10. Articles without available full text were excluded, as were conference abstracts, review articles, case reports, editorial reports, and letters to the editor. Each study was evaluated for eligibility by two reviewers (L.PC. and S.YT.), and study characteristics and outcomes for all selected studies were extracted. A third reviewer (T.J.) meticulously verified the accuracy of the extracted data. The items of these studies that might potentially be related to the outcomes were extracted as follows: first author, publication year, participant characteristics, injection time and dose of SPIO, number of detected patients and positive patients (patients with metastatic sentinel lymph nodes), number of detected sentinel lymph nodes and positive sentinel lymph nodes (metastatic sentinel lymph nodes), as well as other baseline characteristics. When trials had a multifactorial design comparing multiple invention groups, we extracted data and assigned them to the relevant intervention group. We treated each intervention group independently within our analysis [18,19].
All extracted data were tabulated and presented as means and percentages. The R package (version 4.1.3) was used for the quantitative analyses. For (positive) SLN identification, the Cochran-Mantel-Haenszel method was used to test for an association between the two techniques. A similar method was used for detected patients and patients with positive SLNs, across all studies.
We assessed heterogeneity between studies statistically with the I 2 statistic for inconsistency. We used forest plots to estimate the statistical heterogeneity among studies and subgroup analyses. We used I 2 levels of 25%, 50%, and 75% to represent low, moderate, and high heterogeneity [20]. A fixed-or random-effects model was used to account for pooled relative risk (RR) according to their respective heterogeneity (I 2 < 50%, fixed-effects models; I 2 > 50%, random-effects models). The probability for each variable was computed and the RR with a 95 percent confidence interval (CI) was calculated for binary data variables. We considered the comparison significant if p < 0.05 (all tests were two-tailed).
Subgroup analysis, based on the injection dose of the magnetic tracer and the selection of the standard method (RI + BD, RI, or RI ± BD), was also performed to analyze the related efficacy. With a similar initial concentration (Table S1), the dosage-related subgroup analyses were stratified through baseline injection volumes of 0.5 mL, 1.0 mL, 1.5 mL, or 2.0 mL. Of the 19 studies, 1 study was excluded from the subgroup analysis of doses because it randomized participants into three groups according to the usage of doses [21].
The risk of bias in these included non-randomized studies of interventions was assessed using the Risk Of Bias In Non-randomized Studies of Interventions (ROBINS-I) [22]. Seven 'signaling questions' help users judge the risk of bias within each domain. The judgments within each domain carry forward to an overall result across bias domains for the outcome being assessed. Two reviewers (L.PC. and S.YT.) performed the assessments independently. In the case of a disagreement, a consensual decision was counseled by the third author (T.J.).

Description of Studies
The flow of the selection process is shown in Figure 1. Our meta-analysis retrieved 436 articles. After removing 159 duplicates, screening of the 277 titles and abstracts was performed and a total of 235 articles were excluded; only 42 articles remained. Further, nine articles were excluded owing to the invalidity of the reported outcomes for not meeting the inclusion criteria. Moreover, nine articles without parallel controls, four articles compared in two independent cohorts, and two articles compared with the non-standard method were also excluded after the full text assessment. Ultimately, 19 studies from 18 articles were incorporated into the meta-analysis.

Quality Assessment
The quality of the 19 trials was assessed using the Risk Of Bias In Non-rando Studies of Interventions (ROBINS-I). All of the studies (Table S2) were graded as risk of bias in the classification of interventions, bias due to deviation from intend terventions, bias due to missing data, and bias in selection of the reported result. studies were considered at moderate risk of bias in the measurement of outcome study was considered at moderate risk of bias due to confounding, and one stud considered at moderate risk of bias due to selection of the reported result. Overall, were ranked as at moderate risk of bias, while the other 14 studies were rated as lo The main characteristics and specific interventions of the included studies a sented in Table

Quality Assessment
The quality of the 19 trials was assessed using the Risk Of Bias In Non-randomized Studies of Interventions (ROBINS-I). All of the studies (Table S2) were graded as at low risk of bias in the classification of interventions, bias due to deviation from intended interventions, bias due to missing data, and bias in selection of the reported result. Three studies were considered at moderate risk of bias in the measurement of outcomes, one study was considered at moderate risk of bias due to confounding, and one study was considered at moderate risk of bias due to selection of the reported result. Overall, 5 of 19 were ranked as at moderate risk of bias, while the other 14 studies were rated as low risk.
Disadvantages of the magnetic technique were also discussed in some studies [7,12,14,23,[27][28][29]32,34,36]. For instance, the relatively large diameter of the magnetometer's handheld probe would result in larger surgical incisions; the time-consuming frequent balancing of the magnetic baseline level required a correct localization; detection of the local lesion at revision surgery would be challenging for surgeons unfamiliar with the technique, particularly in patients with mastectomies [30]. Moreover, the requirement for the use of plastic alternatives instead of standard surgical retractors, the role of lymphoscintigraphy in successful SLN localization, and the risk of potentially hindering the diagnostic performance of follow-up breast magnetic resonance imaging (MRI) were also reported to be their non-negligible shortcomings [14,21].

Discussion
In this systematic review and meta-analysis, a total of 19 studies on the utilization of SPIO in clinically node-negative breast cancer was included and finally confirmed that SPIO could be an excellent substitution for the standard method whether for identification rates or process improvement. At the same time, we proved lower doses of SPIO have comparable diagnostic accuracy to the standard method. Interestingly, in trials where discordance in detection existed, a greater number of SLNs were identified via the SPIO group compared to the standard group. This significant revelation needs to be noticed because the number of SLNs identified has important implications for the accuracy of the ongoing procedure.
Accordingly, the relationship between the time point choice of administration of the SPIO and the final detection rate has also been explored. Perioperatively [7,12,14,21,[23][24][25][27][28][29][30][31][32][33][34][35][36], one or more days preoperative [26,31,34] injection was comparable to the standard technique in the detection rate. Additionally, higher detection rates than the traditional technique were found if SPIO was injected 1-28 days ahead of surgery compared to that administered on the day of surgery [26,38]. SPIO showed no significant difference in the detection rates between subareolar and peritumoral injections [39]. In another study, a lower SPIO volume injected up to 7 days before the operation has comparable efficacy to the higher SPIO dose (2.0 mL) group and the standard method group for SLN detection [34]. A peritumoral injection and a smaller SPIO dose might also be helpful for addressing the concern of postoperative MRI artifacts [40]. The present results provide convincing evidence that not only a lowered dose but also a flexible injection timeframe in the preoperative period might be combinatorially adopted to enhance the detection rate and SLN retrieval.
SPIO particles do not require any special storage and there are no radiation exposure risks, neither for healthcare personnel nor for the patient. Therefore, non-metal instruments might be adopted to prevent interference between the metal and detection probe. This may present a challenge in obese patients because additional instruments are required to facilitate the dissection of the deep axilla. In addition, a heavier cost burden owing to the equipment of nonmagnetic surgical instruments is also one of their drawbacks, but this could be offset by expense reduction from streamlining processes. Additionally, sparing sentinel lymph node dissection (SLND) procedures were applicable in women with a susceptibility axillary status [25,36,38]. In addition, preoperative lymphoscintigraphy before SLNB was inconsistently used [1,2]. An equally effective outcome was observed in a recent prospective, multicenter, randomized phase III trial, followed by a simplified preoperative workflow and reduced cost, irrespective of the preoperative lymphoscintigraphy results [41].
Another concern is that the depth that the magnetometer can reach is noncomparable to that of the gamma probe, possibly due to the disparate probe size and differential capacity, which can influence the identification of the deeper nodes [24,42]. To this end, novel magnetometry has advanced the appropriate size, excellent detection sensitivity, and good attenuation of noise, as well as the high spatial sensitivity (sufficient depth resolution) to fulfill the pressing clinical requirement [43][44][45]. Moreover, owing to the interference, magnetic nanoparticles should not be used in patients who need MRI to diagnose occult lesions, assess treatment response, or undergo surveillance [40,46]. Of note, the remnants of such nanoparticles would diminish over time and the image quality could be further improved by modifying the sequences [47]. Furthermore, the impact of the differently administered dose of SPIO on skin staining and MRI artifacts is currently studied as well (ISRCTN85167182).
Safety issues regarding the application of SPIO must also be carefully considered. No severe allergic reactions were reported in any of the currently published trials; this can partly be explained by the exclusion of patients with hypersensitivity to iron or dextran compounds and those with pacemakers or metal implants. Dermopigmentation is the most frequently reported complication, with a rate of up to 20% in the injection site, similar to what is found after the injection of blue dye [23,28,48,49]. Skin staining after SPIO injection is a concern as well, even though several reports showed that most patients did not consider it an upsetting problem [21,25]. As reported elsewhere, a deeper injection would be instrumental in reducing the incidence of discoloration [29]. In addition, in a large-scale study, no toxicity in radiotherapy or chemotherapy after SLNB via SPIO was observed [50]. Even though encouraging results have been observed regarding the feasibility of the magnetic technique, further studies with a larger sample size, prospective nature design, as well as rigorous methods of outcome ascertainment, should be warranted before wide clinical implementation [16].
Some limitations also inevitably existed, such as variable blue dyes in the standard method and different cut-off points in applying the SPIO signal to detect the SLNs. However, as clinicopathologic reports were not consistently reported across the studies, these parameters were not included or incorporated in the final comparison. Further, the patientreported outcomes measures (PROMs), an integral component of benefit-risk assessments in the evaluation of new treatment regimens, were generally neglected in most of the studies; this may miss some information and cause some biases in drawing the conclusions. Therefore, randomized trials with a larger sample size comparing the SPIO technique to the standard tracers on locoregional recurrence and survival benefit are desperately needed in the future to advance this technique to benefit the patients who really need it.

Conclusions
Overall, by pooling the trial data, this meta-analysis provides increased evidence that SPIO could be an alternative method to the standard method for axillary node mapping in breast cancer and suggests generalizability of the technique to a wider population.