Phyllodes tumor (PT) of the breast is a rare fibroepithelial neoplasm, accounting for 0.3% to 1% of all breast tumors and 2% to 3% of all fibroepithelial breast lesions (1, 2). PTs are classified as benign (Grade 1), borderline (Grade 2), or malignant (Grade 3) by the World Health Organization (WHO), according to histological features such as mitotic activity, the degree of stromal cellular atypia, infiltrative tumor margins, and stromal overgrowth (3). PT may contain foci with benign, borderline, and malignant features intermingled within the same neoplasm, making careful gross examination and histologic sampling particularly important. Therefore, given PT’s histologic heterogeneity and its propensity to rapidly grow, excision is required to accurately classify, grade, and treat.

The majority of PTs behave in a benign fashion, with the risk of local recurrence (LR) ranging from 17% in benign PT to 27% in malignant PT (4). Several studies questioned the necessity for large margins in benign PT, suggesting that surgical margins less than 1 cm are sufficient (5–8). Based on retrospective data, the National Comprehensive Cancer Network (NCCN) recommends wide local excision (WLE) with a 1-cm margin or more for borderline/malignant PT, but excisional biopsy for benign PT is acceptable (9).

Recently, many studies have shown that complete removal of benign PT using percutaneous excision is feasible (10–14). With this procedure, PTs are removed sample by sample in a piecemeal fashion using 7- to 11-G vacuum-assisted needles under ultrasound guidance (US-VAE). It was postulated that a “wait and watch” approach after vacuum excision (without evident clear margins) could be a safe alternative to local surgical excision with low LR (14, 15). Indeed, complete percutaneous removal would safely exclude a malignant diagnosis and may result in acceptable local control for benign PT.

The present study evaluates whether the LR after US-VAE is different from that observed after surgical excision to determine if US-VAE is a safe minimally invasive option for benign PT treatment.

A total of 1,254 references published between January 1, 1992, and January 10, 2022, were identified. Reviews, conference abstracts, editorials, commentaries, letters, and case reports were also excluded (n = 697). Titles and abstracts were examined to remove articles that were not related to the topic (n = 546). The full texts of potentially relevant articles (n = 11) were retrieved, and the eligibility criteria were evaluated. Six articles were excluded: five did not present the comparator group (surgery) and/or presented incomplete data on recurrence rates following surgery, and one was published in a Chinese periodical that was not available. Finally, all five articles selected were observational cohorts: four retrospective and one prospective ( Figure 1 ). Four papers were from Asia: three from China and one from Korea. One paper was from the USA ( Table 1 ).

Therefore, the meta-analysis comprised the pooling of five studies, compiling a total of 778 participants with benign PT treated by excisional surgery or US-VAE between 2001 and 2020. Of these patients, 439 (56.4%) had local surgical excision, and 339 (43.6%) patients had US-VAE. The median age of patients in the five studies ranged from 33.7 to 39 years; the median T size ranged from 1.5 cm to 3.0 cm, and the median follow-up ranged from 12 months to 46 months ( Table 1 ).

Three studies (Chao et al., 2020, Vargas et al., 2006, and Ji et al., 2022) (11, 12, 19) did not present the Breast Imaging-Reporting and Data System (BI-RADS) classification of the lesions. Ouyang et al. (2016) (14) reported that most patients had BI-RADS 4 lesions in the surgery group (58.1%), and Kim et al. (2016) (20) reported 57.1%. In the US-VAE group, most lesions were classified as BI-RADS 2–3 (58.3%) by Ouyang et al. (2016) (14) and 50% in the study by Kim et al. (2016) (20). Only one study presented the BI-RADS classification of the relapse (Kim et al., 2016) (20). All recurrences of this study occurred in the surgery group and were classified as BI-RADS 4.

In the US-VAE group (n = 339), local recurrence was observed in 41 (12.1%) cases, and in the surgery group (n = 439), 34 (7.7%) cases. There was no significant difference in benign PT local recurrence regardless of the procedure performed (OR = 1.30; p = 0.29) ( Figure 2 ).

There are no randomized data, nor are there ever likely to be, comparing US-VAE, margins < 1 cm, and margins > 1 cm for surgical treatment of benign PT. Our meta-analysis supports US-VAE as an alternative approach to surgery for benign PT, as there is no significant difference in LR after resection by either surgical excision or US-VAE. The current diagnosis and management of PT are based on observational data (1–9). Pathologists find it challenging to exclude malignant phyllodes on core needle biopsy (CNB)/vacuum-assisted biopsy (VAB) sampling and tend to classify them as fibroepithelial lesions. Current practice consists of excisional surgery with clear margins, but not 2 cm wide, as most of these cases are fibroadenoma or benign PT (9). There are emerging data supporting the so-called “wait and watch” approach that consists of follow-up and observation after surgical excision with positive margins or a US-VAE (10–15, 21), and our meta-analysis corroborates this.

The usual clinical scenario is a mass, palpable or not, seen on ultrasound that is submitted to CNB or VAB, yielding an inconclusive result, usually a cellular fibroepithelial lesion (22, 23). These lesions are classified as lesions of uncertain malignant potential (B3 lesions) and were historically always excised. A large fraction (up to 69%) of these lesions turn out to be cellular fibroadenomas (22, 23). Only a very small proportion (up to 2%) of lesions diagnosed as cellular fibroepithelial lesions are malignant PT in the excision specimen (22–24), although the literature shows a wide variation (16% to 76%) of the “upgrade rate” to benign PT on excision (23–26). Another common clinical scenario is a mass, palpable or not, seen on ultrasound that had been submitted to CNB/VAB and was diagnosed as fibroadenoma but shows significant growth over time, raising the possibility of a misdiagnosed PT.

Some or even most of the benign PTs occasionally found in breast cancer screening are round, circumscribed, non-palpable masses categorized as American College of Radiology (ACR) BI-RADS 3 and submitted to follow-up without any percutaneous diagnostic procedure. However, they grow over time, as a rule, and then will be indicated for investigation, and the previously described clinical–pathological situations arise.

In current practice, to exclude malignancy and treat benign PT, the next step in the scenarios above is surgical excision. Most of these excisions result in the diagnosis of benign breast tissue or cellular fibroadenomas (22–26) and could be avoided if a conclusive percutaneous diagnosis with US-VAE was achieved. This analysis shows that even when after US-VAE a benign PT is diagnosed, many surgical excisions could still be avoided using a “wait and watch” approach (27) ( Figure 4 ), and borderline/malignant PT would not be missed ( Figure 5 ). Based on this, we propose a new pathway for lesions submitted to CNB/VAB with a diagnosis of cellular fibroepithelial lesion or benign PT and for lesions with a diagnostic CNB/VAB of fibroadenoma but showing significant growth over time, raising the possibility of PT ( Figure 6 ).

Surgical management of PT is a controversial subject in the literature (2). Wide local excision with minimal 10-mm safe margins is currently being suggested for all subtypes of PT (2). The NCCN guideline (2) recommends wide local excision with the intention of obtaining margins of 1 cm or more for each PT grade. However, their supporting evidence came from a retrospective study that was limited by a small sample size at a single institution (28), and they have recommended that there is no need for re-excision if free margins are achieved for benign PT. Benign PT should be managed differently than borderline/malignant PT (21). There is evidence suggesting that margins < 10 mm or even positive margins do not affect LR for benign PT (21, 29–31). Indeed, a recently published meta-analysis demonstrated that a positive surgical margin was significantly associated with a higher LR risk for malignant PTs, but only a tendency for an increase in the LR risk was observed for benign and borderline PTs (21). In the case of positive margins, there is still the possibility of no re-excision and follow-up: the so-called “wait and watch” strategy (9, 21, 27, 32–34).

There is no relation between margins and distant metastasis for PT, irrespective of grade (benign, borderline, and malignant). Therefore, it is undeniable that breast conservative surgery is the surgical treatment of choice for all subtypes of PT provided that the excised breast volume due to the tumor size does not induce breast deformity (30, 35). Especially for benign PT, a more conservative approach would be appropriate. Our meta-analysis supports the assumption that the proposed pathway is safe.

The upper tumor size limit to indicate a US-VAE for benign PT, and therefore for masses with a CNB/VAB diagnostic of cellular fibroepithelial lesions or benign PT, is controversial and could not be precisely addressed by this meta-analysis. In one study (Chao et al., 2020) (11), tumor size was presented in ranges (2 cm, >2 cm to 5 cm, and >5 cm), and 59% of cases were classified as measuring >2 cm to 5 cm. The mean tumor size at recurrence was not presented in this study. Ji et al. (2022) (12) also presented the tumor size by ranges, with most cases (63.7%) classified as measuring >2 cm to 5 cm. The median tumor size was 2.5 cm between the groups of benign phyllodes tumors treated by US-VAE or surgery. Vargas et al. (2006) (19) presented only the overall mean tumor size (1.57 ± 0.61 cm; 0.3–4.0), which did not differ between the groups (19) ( Table 1 ). In line with these results, if US-VAE complete resection of the visible lesion is achieved, based on the findings of this meta-analysis, further surgical resection seems unnecessary. These findings reinforce recommendations of The Third International Consensus Conference on lesions of uncertain malignant potential in the breast (B3 lesions) that have recognized follow-up as an option for benign PT submitted to US-VAE (36).

There are some limitations to our meta-analysis. Only two studies reported median tumor size in the two groups: 1.7 cm for US-VAE and 3.0 cm for surgery (Ouyang et al., 2016) (14) and 1.6 cm for US-VAE and 3.0 cm for surgery (Kim et al., 2016) (20). Although the median tumor size was smaller in the US-VAE group than in the surgery group, there were no associations between tumor size and local recurrence. Still, the studies are all non-randomized cohorts, and therefore, there is the potential for selection bias in the cases having US-VAE versus those having surgery. The selection criteria to indicate US-VAE or surgery in each of the five institutions of the studies were not completely outlined. US-VAE is a relatively new and not well-established approach. It is possible that some kind of selection bias favored the US-VAE arm in the studies.

Regarding the number of patients and events, this is the largest series (the only meta-analysis) that has addressed US-VAE × surgery for benign PT LR. Although not statistically significant, the LR was higher in the US-VAE group. There was a trend favoring surgery in terms of LR, but there is no guarantee that this is going to be confirmed with a larger cohort. Prospective registration of long-term outcomes for centers that will adopt US-VAE as an alternative to surgery for cellular fibroepithelial lesions and benign PT is recommended.

In addition, surgical procedures are not described in detail. Two studies (Chao et al., 2020 and Ji et al., 2022) (11, 12) described tumorectomy as the excision of the nodule with a surgical margin < 1 cm, wide local excision with a margin > 1 cm, and conservative surgery and mastectomy with a margin ≥ 1 cm. In three studies (Kim et al., 2016; Ouyang et al., 2016; Vargas et al., 2006) (14, 19, 20), the type of surgery performed was not specified. With the available data, it is not possible to analyze the relation of margin width and recurrence between US-VAE and surgery.

There is uncertainty on the pathological diagnostic accuracy of complete US-VAE mass resection for cellular fibroepithelial lesions and PT. PT is classified into three grades according to the WHO: benign, borderline, and malignant. The histological features that are used to classify PT are stromal cellularity, stromal atypia, mitosis, stromal overgrowth, and tumor margin (3). These are also histological features related to LR (21). Three of these could have impaired evaluation. Tumor margin (well-defined, well-defined or focal infiltrative, or infiltrative) could be underestimated, nevertheless not probable to occur. Histological core samples of complete US-VAE provide an evaluation of the borders of the PT, which consists of PT and normal breast tissue, so the microscopic invasion is still possible to evaluate ( Figures 4 , 5 ). Stromal overgrowth, defined by the absence of epithelial elements in one low-power microscopic field containing only stroma (3), could be underestimated or overestimated because vacuum core samples are smaller than a low-power field. Stromal cellularity also could be underestimated or overestimated because of the fragmentation. There would be some challenges to distinguishing between borderline and malignant PT, although benign PT is the low spectrum, and there would be no challenge to its diagnostic ( Figure 4 ).

Furthermore, needle sizes between studies were different. One study used 7-gauge needles (G), two used 11 G or 8 G, and one used 8 G, and one study did not specify. In Vargas et al. (2006) (19), the 8-G needle was used in 169 patients, and the 11-G needle was used in 41 patients. One study (Ouyang et al., 2016) (14) used only 8-G needles. It was not possible to determine the relation between the needle gauge and LR. Probably, the gauge of the needle used does not matter if complete excision is achieved.

Finally, the recurrence intervals between the US-VAE and surgery groups described in Table 1 are in general relatively short and not well described. The study by Chao et al. (2020) (11) presents only the mean overall recurrence interval. It does not differentiate it by groups. Ji et al. (2022) (12) presented only the mean overall recurrence interval (36 months) and did not differentiate it by groups. It is important to realize that in a recent meta-analysis evaluating LR of PT, the median time to recurrence was longer than 24 months in nine studies and shorter than 24 months in eight studies; one of the studies presented a median local recurrence time of 6 months (21). In our meta-analysis, the median follow-up time from the studies ranged from 12 to 48.6 months, and the mean time to recurrence ranged from 17.4 to 36 months. In one of the studies, the follow-up time was just 12 months but showed no recurrence at all in both the US-VAE and surgery groups. We performed the analysis without it and found no difference in the original results ( Supplementary Material ). In addition to this limitation, our results are still of value once the mean time to recurrence ranging from 17.4 to 36 months is in accordance with the literature (21).

In summary, there are no randomized data, nor are there ever likely to be, comparing US-VAE, margins < 1 cm, and margins > 1 cm for surgical treatment of benign PT. Based on these meta-analyses and other observational data, US-VAE appears to be as safe as surgery for the treatment of benign phyllodes tumors, with no difference in local recurrence, and should be an option in the era of personalized treatment. Prospective registration of long-term outcomes is, however, important to further determine the applicability of US-VAE in specific subgroups. These findings are not applicable to borderline or malignant PT, and surgery with at least 1 cm of margin width is still the treatment of choice.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

MG: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Software, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. BAC: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Software, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. HC: Writing – review & editing, Writing – original draft. HS: Writing – review & editing. EP: Writing – review & editing. NS: Writing – review & editing. RM: Writing – review & editing. SM: Writing – review & editing. PD: Writing – review & editing. FC: Writing – review & editing. GG: Writing – review & editing. ADS: Writing – review & editing. WA: Writing – review & editing. JA: Writing – review & editing. CL: Writing – review & editing. ALSF: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Software, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization.