Introduction

The rationale for identifying the sentinel lymph node (SLN), namely the lymph node directly draining the primary tumor, relies on the low likelihood of the presence of cancer cells in the subsequent lymph nodes in the case of a non-metastatic SLN [1]. SLN biopsy (SLNB) is indicated as the gold standard technique for the axillary staging of patients with breast cancer and no clinical evidence of metastatic nodes [2,3,4]. SLNB does not increase the risk of axillary recurrence in patients with breast cancer and reduces the risk of lymphedema compared to complete lymph node dissection (CLND) [5, 6].

SLNB reflects the status of the axillary cavity in over 97% of patients with breast cancer [7]. Radionuclide localization of SLN using 99mTc-labeled colloids in patients with breast cancer is a well-established procedure [8]. Nowadays, planar lymphoscintigraphy (PL) is currently a routine, simple and reliable procedure, performed in most nuclear medicine departments, for the identification of SLN and lymphatic disorders [9, 10]. The inclusion of nuclear medicine procedures for the detection of SLN in the diagnostic workup reduces the false-negative rate of SLNB in patients with breast cancer evaluated at first presentation [11]. Over the last decades, the use of single-photon emission computed tomography/computed tomography (SPECT/CT) has gained wider diffusion in nuclear medicine departments. Indeed, the use of hybrid scanners, providing complementary scintigraphic and morphological data, enables nuclear medicine physicians to offer more accurate information regarding the SLN (e.g. location, number, and surrounding anatomical structures) to the surgeons compared to PL, according to the results of a previous meta-analysis by our group involving patients with melanoma [12].

Whereas an overall superior SLN detection rate (DR) has been reported for SPECT/CT compared to PL in patients with cervical cancer [13] and melanoma [14], such evidence has not been systematically collected for patients with breast cancer.

The aim of this meta-analysis was to perform a head-to-head comparison of the DR of PL and SPECT/CT in patients with breast cancer. Furthermore, as secondary aims, we assessed whether there is a significant difference in the number of detected SLNs, and a significant proportion of patients with additional detected SLNs based on SPECT/CT rather than PL findings or vice versa. Finally, we assessed the ratio of surgical procedures (SLN harvestings) for which SPECT/CT proved a beneficial impact, according to the surgical team.

Materials and methods

The meta-analysis was conducted following the PRISMA guidelines (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) [15]. Before starting the literature search, a protocol was developed to define the research question, search methods, inclusion criteria, quality assessment, data extraction, and statistical analysis. The protocol was registered in the International Prospective Register of Systematic Reviews, Prospero, (www.crd.york.ac.uk/prospero/;protocol CRD42022307723).

Literature search and inclusion criteria

PubMed/MEDLINE and Scopus databases were interrogated independently by two researchers to retrieve prospective or retrospective single or multicenter studies, carrying out PL and SPECT/CT with 99mTc-labeled colloids in patients with primary breast cancer at initial diagnosis before the surgical staging of the axilla.

For our primary outcome (comparison of DRs), we selected articles reporting both the DR of PL and SPECT/CT for SLN (at least 1 lymph node per patient). For our secondary outcomes, we selected articles reporting information on (1) the number of SLNs detected by SPECT/CT and PL, (2) the number of patients with additional SLNs detected by SPECT/CT and/or PL, (3) the ratio of surgical SLN harvestings influenced by SPECT/CT.

The search string was designed to capture the concepts of breast cancer, SLN, SPECT/CT, and PL within the title and article abstracts; for PubMed, the search string was Breast AND (“Sentinel Lymph Node”[Mesh] OR sentinel) AND (“Single Photon Emission Computed Tomography Computed Tomography”[Mesh] OR SPECT). For Scopus, the search string was Breast AND (“Sentinel Lymph Node” OR sentinel) AND (“Single Photon Emission Computed Tomography Computed Tomography” OR SPECT).

No date limit or language restriction was applied. The literature search was updated until March 31st, 2022. All identified references were exported to a reference management software (Endnote v. X7.5, Clarivate Analytics).

Study selection

Two investigators independently screened the titles and abstracts of the records retrieved by the search strings. Only original articles were selected. For each outcome of the present meta-analysis, articles from the same author with the risk of patients’ overlap were also excluded, selecting only the study with the largest number of patients. Duplicates were identified in Endnote and deleted.

After excluding duplicates and non-original articles, the full text of the remaining articles was retrieved to verify the inclusion criteria for this meta-analysis: (1) a study cohort or a subset of a minimum of 10 patients with breast cancer at initial staging undergoing both SPECT/CT and PL in the same day for the identification of the SLN before surgery; (2) injection of 99mTc-nanocolloids; (3) no evidence of other malignancies.

Articles in languages other than English had been planned for translation into English by native speakers before performing the meta-analysis. The references of the retrieved articles were also screened for eventually retrieving additional studies.

Data extraction

Data of all included studies in the meta-analysis were independently extracted by two researchers and any disagreement was resolved in a consensus meeting. Bibliographical and technical data extracted from the articles included: authors, publication year, country, journal, number of patients, sex, and age (mean and range).

For each article, the following data were also retrieved for statistical analysis: the absolute number of patients with at least 1 SLN depicted by SPECT/CT and/or PL, the total number of SLNs detected by SPECT/CT and/or PL, the number of patients with additional SLNs detected by SPECT/CT or PL, the number of patients evaluated for the assessment of the impact of SPECT and PL on the surgical procedure and the number of procedure influenced by the nuclear medicine examination according to the surgeons.

Methodological quality assessment

The methodological quality of the studies was assessed by two investigators using version 2 of the “Quality Assessment of Diagnostic Accuracy Studies” tool (QUADAS-2) [16], which comprises four domains: patient selection, index test, reference standard, flow and timing. The concerns about the risk of bias or applicability were described as low, high, or unclear.

Statistical analysis

Statistical analysis was carried out using MedCalc Statistical Software version 19.1.3 (MedCalc Software, Ostend, Belgium; https://www.medcalc.org; 2020). Publication bias was assessed by visual inspection of funnel plots. The I2 statistic was used to measure the degree of inconsistency across the studies, with I2 values of 25%, 50%, and 75% representing thresholds for low, moderate, and high heterogeneity. Interpretation of heterogeneity was carried out at a significance level of p = 0.05. A random-effects model was used for statistical pooling.

DR was defined based on the detection of at least one SLN in a single patient. Overall pooled DRs were calculated for SPECT/CT and PL on a per-patient-based analysis and presented using forest plots. To assess any statistically significant difference between the two pooled DRs of SPECT/CT and PL, studies were pooled on their odds ratios (ORs) with an inverse variance-weighted random effects model. Pooled data were presented with 95% confidence interval values (95% CI). A statistical difference of pooled DR among SPECT/CT and PL was present if there was no overlap among the 95% CI values.

The number of SLNs detected by SPECT/CT and PL were compared by pooling the ORs with an inverse variance-weighted random effects model. If the value 1 was not within the 95% CI, then the pooled OR is statistically significant at the 5% level (p < 0.05).

The weighted proportion of patients with additional SLNs detected by each technique compared to the other one, and the impact of SPECT/CT on surgery were pooled across the studies and presented in the form of pooled percentages on a per-patient analysis.

Results

Literature search and eligibility assessment

The comprehensive computer literature search from PubMed/MEDLINE and Scopus databases revealed 362 articles (Fig. 1). One-hundred thirty items were duplicates and excluded leading to 232 articles. After excluding non-original articles (n = 110), further 102 abstracts were excluded because they were not in the field of interest of the meta-analysis. The full text of the remaining 20 studies was searched; for two studies, the full text was not available despite contacting the corresponding authors. The characteristics of the retrieved 18 studies [17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34] are presented in Table 1, whereas methodological information concerning the acquisition of PL and SPECT/CT in the studies is summarized in Table 2.

Fig. 1
figure 1

Flow diagram of the literature search

Full size image
Table 1 Characteristics of the eighteen studies selected for the meta-analysis
Full size table
Table 2 Methodological information of the eighteen studies selected for the meta-analysis
Full size table

The risk of bias for the studies included in the meta-analysis was scored low using the QUADAS-2. No publication bias was detected (Fig. 2).

Fig. 2
figure 2

QUADAS-2 results

Full size image

Sixteen studies, with a total number of 2693 patients, were available for the calculation of the pooled DR of SPECT/CT and PL.

Twelve studies were eligible for the comparison of absolute number of SLNs detected by the two techniques.

Thirteen studies were eligible for calculation of proportion of patients with additional SLNs in one of the two techniques.

Four studies (1427 patients) were eligible for the assessment of the average percentage of patients for whom surgical management was influence by one of the two techniques.

Detection rate of SPECT/CT and PL

In a per-patient analysis, the pooled DR for SLN of SPECT/CT was 92.11% [95% confidence interval (95% CI) 89.32–94.50%], whereas the DR for SLN of PL was 85.12% (95% CI 80.58–89.15%) (Figs. 3 and 4, respectively).

Fig. 3
figure 3

Forest plot of the DR for the SLN of SPECT/CT

Full size image
Fig. 4
figure 4

Forest plot of the DR for the SLN of PL

Full size image

The DR rate of SPECT/CT for SLN ranged from 77.78% to 100% across the studies. The DR rate of PL for SLN ranged from 59.70 to 100%. A marked statistical heterogeneity was found for SPECT/CT (I2 = 78.26%) and PL (I2 = 86.76%). A significant difference between the DRs was found with a pooled OR of 1.96 (95% CI 1.51–2.55) in favor of SPECT/CT (I2 = 33.66%).

Comparison of the number of SLNs detected by SPECT/CT and PL

SPECT/CT depicted a higher number (3978 vs. 3321) of SLNs compared to PL in 979 patients, with a statistically significant OR of 1.22 (95% CI 1.14–1.32). No study reported a larger number of SLNs depicted by PL compared to SPECT/CT.

Proportion of patients with additional SLNs detected by SPECT/CT or PL

Taking into account a total sample size of 2485 patients, the pooled proportion of patients in whom SPECT/CT depicted additional SLNs compared to PL was 18.88% (95% CI 11.72–27.27%).

The proportion of patients with additional SLNs detected by PL but missed by SPECT/CT was not significant (0.82%).

Impact on surgery of SPECT/CT

The pooled percentage of cases influenced by the use of SPECT/CT according to surgeons (4 studies enrolling 1427 patients) was 23.98% (95% CI 11.34–39.53%), whereas the corresponding proportion of case influenced by PL was 1.5% (95% CI 0.29–3.90%).

Discussion

Radioisotope imaging has a lower false-negative rate than blue dye and there is no significant difference between indocyanine green (ICG) and radioisotope imaging for the SLN detection, according to a recent meta-analysis (a total of 30 studies, including 4,216 SLN procedures), which, nevertheless, did not analyze the impact of SPECT/CT in the detection performance [35].

In this meta-analysis, we focused on articles comparing SPECT/CT and PL in the same patients at initial staging rather including also studies with parallel data collection of SPECT/CT and PL, in keeping with our previous experience [14]. The reason of our choice is that head-to-head comparison provides a more accurate estimate of the outcome measures compared to matched-pair comparison [36]. We limited our analysis to patients at staging, because in case of breast cancer recurrence and previous axillary lymph node dissection (ALND), the repeat sentinel node biopsy has a significantly lower rate of harvesting and a much more aberrant lymphatic [37].

Very high rates of successful SLN detection in patients with breast cancer have been reported with either PL or SPECT/CT [24]. Nevertheless, the use of SPECT/CT has been encouraged by several authors due to substantial advantages over PL, including higher DR, better spatial resolution, more precise anatomical localization of the SLN [38] and efficient attenuation correction through the exploiting the CT data [13, 39, 40]. Conversely, the use of SPECT/CT increases the acquisition time and the radiation dose compared to PL, potentially reducing the patient workflow and bringing additional costs [41].

Our meta-analysis documented a superior DR for SPECT/CT compared to PL and a larger number of SLNs detected by SPECT/CT compared to PL. Higher DR and a larger number of SLNs identified by SPECT/CT in comparison with PL may also determine a meaningful impact on surgical decision-making. Nevertheless, the preoperative use of SPECT/CT for the identification of SLNs is not important only for the additional number of SLNs but also for the capability of providing anatomical information [42]. Indeed, SPECT/CT may also localize unspecific hot spots that could be mistaken as additional SLNs using PL only [43], for example in case of cutaneous contamination, skin fold, propagation from the injection site or leakage from the wire tract [27]. We found a 23.98% of change in surgical approaches in patients with breast cancer. A more precise localization of SLNs may lead to a more precise surgical procedure (due to a change in the location, size and accuracy of the incision), facilitating the surgical planning, reducing the morbidity, the duration of surgical operations and costs [44].

From our analysis, it can be observed that heterogeneity indexes are high either for SPECT/CT (I2 = 78.26%) or PL (I2 = 86.76%), which is in contrast with our previous results in melanoma patients [14], which suggested higher repeatability for SPECT compared PL, as highlighted by the lower heterogeneity index (I2) obtained for SPECT/CT (I2 = 62.45% vs. 78.96%).

van der Ploeg and coworkers suggested three main indications to perform SPECT/CT: (1) inconclusive PL due to unusual lymphatic drainage pattern, (2) difficulty in the interpretation of a lymphatic pattern, and (3) nonvisualization of SLN at PL [45], accounting approximately for a third of patients with breast cancer according to their experience [33]. Also, the recent systematic review of Ge et al., suggest to add SPECT/CT to the diagnostic work-up of patients with breast cancer recurrence when PL fails to detect the SLN, since SPECT/CT may increase the chance of detecting the lymph node also outside the axilla [37].

Certain patient characteristics may also lead to add SPEC/CT to the radioisotope mapping. Increased body mass index (BMI) is a major risk factor for the development of severe lymphedema after ALND [46] and, along with breast size, is also an important factor affecting negatively lymphoscintigraphic and intraoperative SLNs detection [47, 48]. On the other hand, Lerman et al. demonstrated the superior performance of SPECT/CT in identifying SLNs in overweight and obese breast cancer patients, even in case of intraoperative blue dye technique failure, suggesting the indication to routinely perform SPECT/CT in case of high BMIs [27].

The 2013 joint EANM and SNMMI guidelines do not indicate SPECT or SPECT/CT as mandatory, but as optional or alternate imaging, suggesting its use in the case of nonvisualization of SLN on PL, obese patients, and SLNs outside the axilla. Otherwise, the execution of SPECT/CT may be justified by the difficulty in the interpretation of PL due to unexpected or unusual lymphatic patterns (e.g. multiple sites of drainage or the appearance of the intramammary lymph node chain, SLN in the contralateral axilla, previous breast surgery, the presence of a SLN near the injection area, or suspicion of contamination) [49]. Conversely, the Chinese Society of Breast Surgery in 2021 attributed a level B (weak) strength of recommendation to lymphoscintigraphy as a mapping method of the SLN, not even mentioning SPECT/CT [50].

The current use of PL as the elective method for radioisotope mapping of the SLN relies on its high detection rate [41]. There are still few data to demonstrate that the addition of SPECT/CT improves staging, since only a few studies reported the histologic data of the SLNs depicted by SPECT/CT but missed by PL [33]. Additionally, there are still no sufficient data to confirm that a higher number of SLN identified by SPECT/CT and removed result in an improvement of control disease. Nevertheless, SPECT/CT is strongly recommended for selected indications, especially when PL fails to detect the SLN, in case of abnormal lymphatic drainage pattern, and for overweight patients.

Some limitations may affect our meta-analysis. The selected studies provided markedly variable sample sizes. Another source of bias may derive from the high heterogeneity of the DR across the studies. Further sources of bias may arise from some differences across the studies including the number of radiotracer injections, and methodology of PL and SPECT/CT execution. As suggestions for further studies, cost-effectiveness analyses should evaluate whether the use of preoperative SPECT/CT compared with PL for SNL detection in breast cancer is associated not only with higher detection of metastatic involvement but also with a significant cost reduction.

Conclusions

The present meta-analysis favors the use of SPECT/CT with 99mTc-labeled colloids over PL in patients with breast cancer for the identification of SLN due to its superior DR. Further advantages of SPECT/CT over PL are an overall larger number of depicted SLNs, a significant proportion of patients with additional SLNs detected by SPECT/CT but missed by PL, and an impact on surgical strategy on a significant percentage of patients. Nonetheless, in institutions where SPECT/CT is not available, PL remains a good option due to its good DR for the SLN on a patient-based analysis.