Mammographically detected spicules associated with malignant breast tumors frequently harbor additional tumor foci

Breast cancer is a heterogeneous disease that can be multifocal in nature, irrespective of tumor size or histology, which poses challenges in its management. Early results on mastectomy specimens showed that only 37% of T1–2 invasive breast cancers were unifocal, while the rest had additional tumor foci.¹ Such additional foci can lead to positive resection margins after breast-conserving surgery (BCS), and thus, an increased risk of reoperation or local recurrence. This highlights the importance of thorough preoperative assessment to identify any additional tumor foci that might not be apparent upon initial examination. Based on our own observations, most occult tumor foci detected after BCS are localized within tumors’ spicules.

A spiculated breast lesion can be defined as a mass or architectural distortion with thin lines radiating from its margins.² Spicules are common in malignant breast lesions yet can be observed in both benign and malignant breast processes.² The exact mechanisms underlying malignant spicule formation beyond the tumor mass are not fully understood, and its clinical significance is unknown. It is speculated that spiculation involves two time-dependent steps: cancer cell invasion initially associated with a desmoplastic stromal reaction, followed by adipose tissue invasion and/or involvement of Cooper’s ligaments by carcinoma tissue.³ Adipose tissue can release hormones and active adipokines that participate in the regulation of the immune response, cancer metabolism, and energy balance.^4,5 Adipocyte–cancer cell crosstalk is important for the desmoplastic reaction, which might contribute to spiculation.⁶ Furthermore, the extracellular matrix is not a passive bystander in cancer progression.⁷ Collagen fibers create pathways that cancer cells can use to invade the environment.⁸ Consequently, spicules are not only connective tissues but may also contain tumor cells.⁸ Since focal invasion begins at the base of the spicule, it has been suggested that tumor cells should be more abundant at the spicule’s base than at its end.⁸

Spiculations pose also a challenge for imaging. Mammographic features cannot reliably differentiate whether the spiculated lesion is benign or malignant, resulting in the need for a biopsy unless the lesion is indisputably postsurgical fibrotic tissue.⁹ Spiculations are difficult to detect in dense breast.³ Although spicules are usually visible on full-field digital mammography (FFDM), summations from normal fibroglandular tissue might hinder the evaluation of spicules on conventional mammograms.⁸ Digital breast tomosynthesis is a three-dimensional imaging modality that eliminates tissue superimposition and provides a clearer view of the lesion and its margins and spicules as well as associated architectural distortions, making it easier to differentiate spicules from the summations of the fibroglandular tissue.⁸ Furthermore, tomosynthesis has recently been shown to be superior to other mammographic imaging modalities for the evaluation of resected specimens.¹⁰

To the best of our knowledge, no studies have been conducted to assess (and no guidelines are available on) the surgical removal of mammographically detected spicules or the frequency of occurrence of tumor foci in spicules. Therefore, this study was designed to evaluate the frequency and nature of additional occult malignant lesions and their relationship with spicule size and quantity, as visualized by specimen tomosynthesis.

A total of 162 invasive lesions from 156 women (mean age: 63.0 ± 10.2 years, range: 33–95 years) fulfilled the inclusion criteria and were evaluated. The patient and tumor characteristics are presented in Table 1, and the mammographic characteristics of the tumors are presented in Table 2.

A summary of the features of the spicules observed in the pre- and intraoperative images is shown in Table 3. There was no statistically significant difference in the presence or distribution of spicules between the pre- and intraoperative images. However, there was a statistically significant difference and a strong positive correlation between the pre- and intraoperative images was found regarding the number and maximum length of the spicules.

The mean pathologically measured distance from the main tumor margin to the additional spicule-associated tumor was 4.3 ± 2.8 mm. Compared to the maximum spicule length measured in the intraoperative images (9.5 ± 5.1 mm), the distance of actual tumor infiltration was much shorter, and the correlation was very weak (P = 0.055, r = 0.234).

In terms of their distance from the main tumor margin, most tumors in the spicules were < 5 mm away (55, 60.4%), some were 5–9 mm away (30, 33.0%), and a small proportion were ≥ 10 mm away (6, 6.6%). The distribution of spicule lengths in the preoperative images was as follows: 38 (39.2%) were < 5 mm in length, 34 (35.1%) were 5–9 mm in length, and 25 (25.8%) were ≥ 10 mm in length. The pathologically and radiologically categorized measurements showed significantly different distributions (P = 0.002) and no agreement (P = 0.66, κ = 0.04). The types of tumors found in the spicules were as follows: no additional tumor (70, 43.2%), invasive only (34, 21.0%), in situ only (31, 19.1%), and both invasive and in situ (27, 16.7%). Representative images are shown in Figures 1–Figures 3.

Figure 1.

A 14 mm spiculated invasive ductal carcinoma (IDC) excised en bloc with spiculations. A tomosynthesis specimen radiogram at the level of the tumor documents the total removal of both the guidewire localized tumor (A; arrow) and the associated spicules (arrowheads). The histopathological analysis (B, H&E; 10 × magnification) showed that all the spicules contained additional tumors: the spicule shown in the red box contained IDC (100 × magnification); one spicule contained ductal carcinoma in situ (DCIS) (blue box, 150 × magnification); other spicules contained both DCIS and IDC in green (100 × magnification) and yellow (150 × magnification) boxes.

Figure 2.

A mammogram of a nonpalpable, spiculated 12 mm invasive ductal carcinoma: a craniocaudal-tomosynthesis view (A) demonstrates spiculations better than full-field digital mammography in mediolateral-projection (B). After breast conserving surgery, a more superficial axial section (C) the invasive additional cancer extends 2.2 mm from the tumor’s margin. In the deeper section (D) additional invasive cancer cells are visible further in the spicule extending up to 2.7 mm from the tumor’s margin highlighting that analyzing only one section of a spicule will underestimate the extent of the disease. (E) A magnified view of the base of the spicule.

Figure 3.

(A) A mediolateraloblique spot magnification view of a 13 mm mainly circumscribed, mixed papillary and invasive ductal carcinoma (arrows) with a few spicules (arrowheads). The histopathologal analysis (B) showed a single solid spicule with multiple foci, likely representing a continuum of low-grade ductal carcinoma in situ (arrows) extending as far as 10.1 mm from the tumor’s edge and close to the resection margin.

Tumors that had spicules containing additional tumor material were statistically larger than those that did not have spicules containing additional tumor material (17.9 ± 11.6 vs. 13.8 ± 6.9 mm, respectively, P = 0.006).

None of the tested biological markers (i.e. ER, PR, HER-2, Ki-67, or grade) showed any significant relationship with the distribution, number, or length of spicules, the presence of additional tumors in spicules, or their distance from the main tumor margin.

Table 4 summarizes the distribution of the tumor subtypes in relation to spiculation and the presence of additional tumors in the spicules. Luminal A and B tumors more frequently presented with spiculation as a dominant feature (χ²= 10.4, P = 0.016), and more additional tumors were found in spiculated masses (χ² = 5.6, P = 0.018) compared to non-spiculated tumors.

The results of this study suggest that the presence of spicule-associated tumor foci is a common finding in breast cancer, irrespective of the length or quantity of the spicules. Furthermore, in this study, the overwhelming majority of these additional foci were within 1 cm of the tumor edge. These results strongly imply that spiculation is not insignificant and should be reported and included in the preoperative surgical plan. To the best of our knowledge, there have been no previous studies on this issue, and no surgical guidelines are available for the management of mammographically detected spiculations.

Even though residual microscopic disease after surgery will most likely be eradicated by post-surgical adjuvant treatments, the presence of positive margins after resection significantly increases the risk of local relapse. Spiculated tumors have been linked to an increased risk of ipsilateral recurrence^3,15, although low-grade tumors are more likely to exhibit a spiculated margin compared to rapidly growing high-grade tumors.¹⁶ In addition, during imaging, the appearance of low-grade tumors is influenced by the presence of tumor- associated fibrous connective tissue (i.e., the desmoplastic reaction).¹⁷ However, Moriuchi et al.³ found no statistically significant differences in the 10-year disease-free or overall survival rates between patients with spiculated and non-spiculated masses.³ Tabar et al.¹⁸ reported that small (1–14 mm) spiculated tumors were associated with excellent long-term survival.

While the routine removal of all spicules during surgery might decrease the risk of positive margins, it will result in larger resection, more complex surgery, potentially inferior aesthetic results, and lower patient satisfaction. Reoperation rates vary significantly among centers and surgeons, with an average rate of 20%.¹⁹ Our results may explain the high frequency of positive margins reported after lumpectomy.²⁰ Cavity shaving has been shown to reduce the positive margin and re-excision rate by half²⁰, and assumingly shaving removes a significant number of peritumoral spiculations. Finnish national guidelines recommend wide local excision of tumors, with a minimum lateral macroscopic margin of 1 cm; however, different centers have reported different reoperation rates, ranging from 8.4 to 13%.^21,22 In our local practice, we strictly adhere to this recommendation, and based on our observations, we routinely recommend the removal of all associated spicules that, on average, are about 1 cm in length. This approach, together with our meticulous preoperative evaluation process, may partly explain our previously reported low reoperation rates of 0.5–2.1%.^10,23

Intraoperative specimen tomosynthesis was previously shown to be superior to both FFDM and dedicated digital specimen radiography system in the evaluation of lesion size, margins, spicules and calcifications.¹⁰ However, in this study, intraoperative specimen tomosynthesis was not significantly superior to extensive preoperative imaging in terms of the visualization and characterization of spicules. Therefore, in our opinion, it can be used to confirm that the detected spicules have been excised. Furthermore, our results indicate that analyzing preoperative FFDM images plus additional images, including spot compression and/or tomosynthesis images, is sufficient to properly plan surgical interventions.

Other studies have reported that mammographic spiculation is a good independent prognostic factor for screening-detected invasive breast cancer.^24,25 These favorable characteristics include ER and PR positivity as well as lower histological grade and Ki67 expression.^24,26 Samaržija et al.⁸ found that longer and thinner spicules were associated with a low Ki-67 index, whereas shorter and thicker spicules were associated with a higher Ki-67 index. In our study population, we did not find any significant relationship with 1) the spicule distribution, number, or length, 2) the presence of tumors in the spicules, or 3) their distance from the main tumor margin.

The mammographically detected spicule length correlated only weakly with actual tumor infiltration, and no correlation was found with any measurable parameter that could help predict the presence or extent of tumor cell infiltration in spicules. Even though the majority of spicule-associated tumors (60.4%) were located less than 5 mm from the tumor edge, additional foci were frequently found outside this range. Therefore, even a margin of ≥ 1 cm has a 6.6% theoretical risk of positive margins.

This study has some limitations. The results should be interpreted with caution, as the spicules were not fully evaluated histologically via serial sectioning. Serial sectioning requires extensive resources in a prospective setting, whereas this study was retrospective in nature. The specimens and samples were handled and prepared according to the routine protocol of the Department of Clinical Pathology. However, this protocol does not include sampling of all spicules, some of which are not macroscopically visible. Therefore, it is likely that not all spicules were histopathologically evaluated or included in the correlation analysis with the imaging findings. Consequently, the number of tumor foci in the spicules was likely underestimated. Furthermore, the tumors examined in this study were relatively small (pT1 75.3% and pT2 23.5%), had less aggressive prognostic features, and were mostly early-stage nonpalpable cancers that required intraoperative specimen assessment. Given that the lesions that had additional tumors were statistically larger than those that did not have additional tumors, it can be assumed that larger or palpable tumors with potentially longer spicules might have more abundant additional disease. Additionally, the specimens examined in this study were obtained from a single institution.

Nevertheless, this exploratory study is the first of its kind and was conducted to verify the clinical value of spicules, which might have practical clinical potential. To reduce bias, consecutive patients were included in this study. Our results clearly highlight the need for future prospective research that includes extensive serial sectioning of the spicules.

In conclusion, the results of this study show that spicules are significant structures in that they frequently harbor tumor foci, which might result in margin positivity and potential reoperation. Additional research is necessary to determine the tumor burden and clinical significance of both spicules and their tumor foci.