Clinical presentation and radiologic imaging findings of phyllodes tumors: benign and borderline/malignant phyllodes tumors

Study population

The data of female patients with histopathologically confirmed phyllodes tumor who had undergone preoperative mammography or ultrasonography or post contrast CT chest at Thammasat University Hospital from 1 June 2011 to 30 Sep 2021 was collected and retrospectively reviewed. The CT chest was performed in some patients due to the huge breast mass that cannot perform mammography and ultrasound.

The inclusion criteria were as follows; (a) patient who underwent breast surgery and received a histologic diagnosis of phyllodes tumor (b) the pre-operative radiological images (digital mammography or ultrasonography or post contrast CT chest) were available on Pictures Archiving and Communications System (PACS) of the Thammasat University Hospital, and (c) the medical records and pathological results were accessible on the information system of the Thammasat University Hospital (EPHIS). The exclusion criteria were as follows; (a) incomplete medical record and (b) inconclusive pathological results such as fibroepithelial tumor.

Of the initial 70 suspicious phyllodes tumors, 10 cases were excluded because of incomplete medical record and unavailable pre-operative images, and 3 cases had inconclusive pathological result as fibroepithelial tumors from core needle biopsy and no other procedure to receive more tissue. Eventually, a total of 57 phyllodes tumors were eligible for our study. The final pathologic results of phyllodes tumors were benign for 43 cases and non-benign (borderline and malignant) for 14 cases.

Imaging technique and processing

The patients were received at least one of the imaging studies as follows;

Clinicopathological evaluation

The patient’s clinical information was collected from electronic medical records consist of age at diagnosis, clinical presentation (palpable mass, breast pain, skin ulcer, or screening), duration of symptom, menstrual status, side of tumor (right or left), location of the tumor, treatment (wide excision or mastectomy), and metastasis (none, lymph node, or distant).

The histopathological data were retrieved from the final histopathological reports which were separated into two groups: benign and non-benign (borderline and malignant) phyllodes tumor. The phyllodes tumor was categorized based on histopathological features, using World Health Organization (WHO) criteria which divided histologic features of phyllodes tumor based on stromal atypia, mitosis, stromal overgrowth, and margin.²

Imaging interpretation

The images (mammography, ultrasonography, or contrast enhanced CT chest) were retrospectively reviewed by two radiologists with 13 years and 12 years of experience in breast imaging. For discrepant results, a final consensus was reached after discussion. The images were randomly selected and blinded to histopathological results.

The images were retrieved retrospectively from PACS. The latest images before undergoing treatment were selected and reviewed. The details of each imaging modality were described as follows;

Data analysis and statistics

The demographic data of the patients and histopathological results of all phyllodes tumors were reported by using the number (percentage) for categorical variables and mean ± SD or median (IQR) and range for continuous variables.

To compare the clinicopathological and imaging characteristics between benign and non-benign phyllodes tumor, we initially used Shapiro-wilk normality test before using Student’s t-test for normally distributed continuous variables and Wilcoxon rank sum test for non-normally distributed continuous variables. The Chi-square test or Fisher’s exact test was used for the categorical variables. All statistical analyses were performed with R program (version 4.1.1, R foundation for Statistical Computing, Vienna, Austria) and statistical significance will be considered as p-value less than 0.05.

With regard to the data analyses, the demographic data and imaging features were discovered to be statistically significant; logistic regression analysis was conducted to calculate the odds ratios (OR) with 95% confidence intervals (95% CI) for predicting non-benign phyllodes tumor probability as compared with benign phyllodes tumor. Odds ratios were contemplated to indicate statistical difference if the 95% CI excluded 1.0. A univariate analysis was first performed to identify any potential predictor variables. The demographic data and imaging features with a p-value <0.05 according to a univariate analysis were included in the multivariate analysis to determine any independent predictors of the non-benign phyllodes tumor. A stepwise selection procedure was used to identify variables with a p-value <0.05.

Patient characteristics and histopathological results

A total of 57 female patients with a mean age of 38.8 years (range, 15-66 years) were included in the study. Forty-three patients were benign, 7 patients were borderline, and 7 patients were malignant subtype. The demographic data, clinical characteristics, and histopathological results of patients are shown in Table 1.²²

There was no statistically significant difference in patient’s age, clinical presentation, duration, side, treatment or metastasis between benign and non-benign groups. The mean ages of patients with benign and non-benign phyllodes tumors were 37.1 years and 43.9 years, respectively. Palpable breast mass was the most common (91.2%) presenting symptom in both groups. Noted that in some patients can had more than one symptom at the time of presentation. The median duration of symptoms prior to diagnosis was 4 months, which ranged from 0.03 month (1 day) to 60 months (5 years). For the patients whose phyllodes tumor incidentally found on screening were not included for statistical analysis. Both sides of breast were equally affected from both benign and non-benign phyllodes tumors.

Most patients (77.2%) from both groups underwent wide excision of the tumor as a treatment. The duration of the first imaging procedure to surgery were ranged from 1 day to 610 days and the median duration was about 110 days. Time interval between imaging procedure and surgery of malignant Phyllodes tumors were 1 day to 31 days. Some patients who were benign or borderline Phylloes tumor, had time interval more than 6 months because of significantly increased size of tumor during imaging follow up. There were 2 distant metastatic patients (14.3%) from the non-benign group which both of them were lung metastasis. None of the non-benign phyllodes tumor patients had lymph node metastasis.

There were statistically significant differences in the menstrual status (p=0.009) and location of the tumor (p=0.027) between benign and non-benign phyllodes tumors. The majority of patients in benign group were pre-menopausal status (79%), while 50% of the patients in non-benign group were menopausal status (p=0.009). Tumor involving entire breast was more presented in patients with non-benign phyllodes tumor (p=0.027).

Mammographic features

Thirty-one patients underwent mammography. The patients who did not undergo mammography due to 1) the tumor size was too large to be performed mammography, and 2) patients were younger than 30 years old, so they were undergoing ultrasonography for the evaluation instead of mammography. However, our study did not show any statistically significant differences in all mammographic features between the two groups as demonstrated in Table 2. Noted that of all 31 patients, 7 patients were excluded from size evaluation because the masses were obscured by surrounding dense breast parenchyma.

Ultrasonographic features

All 57 patients with phyllodes tumor underwent ultrasonography. There was no statistically significant difference in the tumor’s side, shape, border, posterior acoustic features, and cystic area between benign and non-benign groups as demonstrated in Table 3.

There were statistically significant differences in the size (p= 0.038), echogenicity (p= 0.019) and vascularity (p= 0.04) of phyllodes tumors between benign and non-benign groups. Size of the tumors was commonly less than 5 cm in both benign and non-benign groups (30 of 43 [70%] vs 6 of 14 [43%], respectively). Whereas the tumor size more than 10 cm was more common in non-benign than those of benign group (7 of 14 [50%] vs 4 of 43 [9%], respectively). The most common echogenicity of benign phyllodes tumors was hypoechogenicity (Figure 1). Whereas the heterogeneous echogenicity (Figure 2) was higher found in non-benign phyllodes tumor compared to benign phyllodes tumor (8 of 14 [57.1%] vs 8 of 43 [18.6%], respectively). None of the phyllodes tumors had hyperechogenicity on ultrasonographic images. The most common vascularity characteristics in both benign and non-benign phyllodes tumors was presence of internal vascularity (37 of 43 [86%] vs 8 of 14 [57.1%], respectively) (Figure 3).

Figure 1. A 15-year-old girl with a palpable left breast mass for six months.

The ultrasound showed a 3.0x2.3x3.3-cm well-defined lobulated hypoechoic mass with posterior acoustic enhancement at the left 12 o’clock. Wide excision of the mass was done, and the histopathologic result was a benign phyllodes tumor.

Figure 2. A 42-year-old woman with malignant phyllodes tumor.

The ultrasound showed a huge circumscribed heterogeneous echoic mass with internal cystic spaces involving the entire right breast.

Figure 3. A 62-year-old woman with palpable right breast mass for one year.

A. The gray-scale ultrasound showed a large, partially well-defined, heterogeneous hypoechoic mass with small internal cystic spaces involving the outer region of the right breast and B. The color doppler ultrasound showed internal vascularity in the mass. The histopathologic result was a benign phyllodes tumor.

Univariable analysis (Table 4), using the logistic regression to analyze the correlation between demographic data and ultrasonographic features to predict non-benign phyllodes tumor, the results showed that patients with menopausal status (p= 0.008), tumor located in entire breast (p = 0.015), tumor size more than 10 cm (p= 0.029) and heterogeneous echo on ultrasound images (p= 0.009) were significantly related with non-benign phyllodes tumor.

After multivariable analysis (Table 4) with multicollinearity avoidance with stepwise selection procedure, menopausal status (odd ratios, 13.79 [95% CI: 2.31,82.24], p value of 0.04), and presence of vessels in rim (odd ratios, 16.51 [95% CI: 1.59,171.96], p value of 0.019) or absent vascularity on doppler ultrasound (odd ratios, 8.45 [95% CI: 1.03,69.15], p value of 0.047) were significantly increased possibility of non-benign phyllodes tumor. But the location, tumor size and echogenicity showed no statistically significant difference for predicting non-benign phyllodes tumor.

CT features

There were only 5 (8.8%) from 57 phyllodes tumor patients who underwent contrast enhanced CT chest. One (20%) patient was benign and 4 (80%) patients were non-benign phyllodes tumors. Because of the small number of subjects, we would only describe CT findings without statistical analysis as demonstrated in Table 5.

The one benign phyllodes tumor was found in the right breast, while the non-benign phyllodes tumors were found in each side of the breasts equally. The size of phyllodes tumor ranged from 8.9 to 24.5 cm in the longest dimension, which both of the smallest and the largest tumors were found in non-benign group. Only 1 non-benign phyllodes tumor showed microcalcification, while the rest of phyllodes tumors did not present calcification. All of phyllodes tumors were irregular shape, heterogeneous density, involving more than 2 quadrants and located in entire breast. The border of phyllodes tumor was circumscribed in 3 (75%) patients from non-benign group whereas irregular border was found in 1 (100%) patient from benign group and 1 (25%) patient from non-benign group. There were 2 (50%) non-benign phyllodes tumors that showed no enhancement, while the other 2 (50%) non-benign phyllodes tumor had marked enhancement (Figure 4), and 1 (100%) benign phyllodes tumor showed moderate enhancement.

Figure 4. A 37-year-old woman with borderline phyllodes tumor.

A. The pre-contrast CT chest showed an irregularly shaped, heterogeneous hypodense mass involving the entire right breast with skin invasion and B. After contrast administration, the tumor showed marked heterogeneous enhancement.

Limitations

There were some limitations in this study. First, the study was a retrospective design study that may have had a selection bias. Second, this was single-centered study that may cause a relatively small sample size and an unequal number of patients between benign and non-benign phyllodes tumor. Multicentered study with a larger population and data sets may be needed to validate our findings.