Outcomes of breast conserving therapy: Recurrence, imaging findings and histological correlation

Background Breast conserving therapy (BCT) is the mainstay therapy in patients with early breast cancer and selected patients with locally advanced breast cancer. No formal audit has been performed on BCT at our institution. Objectives To determine the incidence and risk factors for ipsilateral breast tumour recurrence (IBTR). Study the imaging features of IBTR. Determine adherence to the proposed annual mammographic surveillance schedule. Method Clinical, radiological and histopathological records of patients who underwent BCT from 01 January 2011 to 31 December 2015 were reviewed. Patients were followed up for at least 5 years. Results Ninety-two patients were included in the study with a mean age of 54.3 years. Eighty of the 92 (87.0%) patients were imaged within 1-year post-BCT. Ipsilateral breast tumour recurrence was 6/92 (6.5%) with mean time to IBTR of 34.4 months. One of the 92 (1.0%) patients had a contralateral metachronous recurrence with no IBTR and 11/92 (12.0%) had distant metastases only. Pathological tumour size and extent (pT2) (68.5%) and pathological lymph node (pN0) (65.2%) were the most common locoregional staging. Infiltrating ductal carcinoma was the most common histological type (88%). Age < 35 years was associated with breast cancer recurrence (p < 0.01). Imaging findings of recurrence were microcalcification (odds ratio [OR]: 4), asymmetric density (OR: 4) and skin thickening (OR: 2.5). Conclusion The occurrence of IBTR following BCT in our unit is acceptable and comparable to local and international units. The accuracy of assessing the post-BCT breast for IBTR is in keeping with international standards. Contribution Improved radiological imaging interpretation of the post-BCT breast.


Introduction
Breast conserving therapy (BCT) is the mainstay therapy and an alternative to mastectomy in patients with early breast cancer (EBC) and is the standard care at our centre, Inkosi Albert Luthuli Central Hospital. 1,2,3,4 Breast conserving therapy offers the advantage of an acceptable cosmetic result; however, the post-treatment breast is difficult to manipulate for imaging due to patient pain threshold limitations and anatomical distortion that impairs breast compressibility, potentially masking an underlying ipsilateral breast tumour recurrence (IBTR). 5,6 Early detection of IBTR is linked with greater survival outcomes. 7 Clinically and radiologically diagnosed IBTR through surveillance screening in asymptomatic patients has a better prognosis than patientdetected recurrences. 7 It is therefore imperative that the imaging findings are accurately assessed for prompt diagnosis and treatment of IBTR. 5,6,7 The recurrence rates post-BCT vary with an acceptable range of 5% -15%, although some studies have reported higher and lower incidences. 8,9,10,11,12,13 Local studies performed at Groote Schuur Hospital in Cape Town showed a 5-year recurrence of 6.8% (median followup of 60 months) 14 with a smaller study conducted in Johannesburg reporting a 5.5% recurrence (median follow-up of 65 months). 15 Newer studies report significantly lower recurrence rates. 8 Ipsilateral breast tumour recurrence leads to an increased risk for distant metastasis and is associated with decreased survival. 16,17 The most frequent site is the original quadrant or the chest wall scar after radical surgery, which accounts for 60% -95% of all cases. 16 Several risk factors have been associated with IBTR. Young age (defined as either < 35 years or < 40 years) is associated with early onset and higher rates of IBTR when compared to older counterparts. 18,19 A positive family history (particularly > 3 first-degree relatives) increases the risk of IBTR and contralateral recurrence. 20 Breast cancer (BRCA) gene mutation 1 and 2 carriers are at risk of IBTR. 21 Pathological margin status is an independent risk factor for IBTR. 22,23,24,25 Triple negative subtype, ductal carcinoma in situ (DCIS) and an increased number of positive lymph nodes (particular N2) are associated with an increased risk for IBTR. 26,27 Noncompliance with, interruptions and delays in initiating adjuvant radiotherapy and omission of a tumour bed boost are significantly associated with poorer outcomes. 12, 28,29 Tumours that fail to demonstrate pathological complete response after neoadjuvant chemotherapy are associated with a higher risk of IBTR. 27 Ultrasonography is sensitive in distinguishing between postoperative complications and benign abnormalities but is inferior to mammography in detecting IBTR. 30 Most IBTRs are described as hypoechoic lesions in contrast to benign lesions which tend to be anechoic or hyperechoic. 30 The sensitivity of mammography in detecting IBTR in a posttreated breast ranges from 55% to 68%; this is due to surgery and radiation-induced changes in the parenchymal pattern. 5,17,31 A small study by Pinksy demonstrated a 97% yield in diagnosing recurrences on mammography with 91% of cases being clinically occult and diagnosed solely on mammography. 32 Mammographic characteristics of IBTR include calcifications at or near the tumour bed which usually resemble the primary tumour and may be pleomorphic, indistinct, coarse heterogeneous, or linear (75%); areas of architectural distortion (3%) and increasing skin oedema. 6,32 Ipsilateral breast tumour recurrence may appear as a mass or asymmetric density of similar mammographic appearance to the primary tumour (p < 0.02). 33 MRI is superior to ultrasound (US) and mammography due to its ability to identify additional disease in the ipsilateral and contralateral breast, detecting 15% and 10% of additional disease, respectively. 34,35 In a study by Lehman et al., the sensitivity and specificity of MRI in the contralateral breast were 91% and 88% respectively with a negative predictive value of 99%. 36 On mammographic and US images, tumour size was significantly underestimated by 14% and 18%, respectively (p < 0.005), while MR imaging showed no significant difference in size compared with that found at pathologic evaluation. 35 In cases where mammography and/ or ultrasonography are indeterminate for IBTR, MRI can reliably distinguish post-BCT scar tissue from IBTR when performed at least 12-18 months after completion of BCT, yielding a sensitivity of 90% -100% and specificity of 83% -93%. 5 Pre-operative MRI also allows better assessment of the extent of a cancer in surgical naïve breasts, leading to better surgical planning (wider excision, extra biopsies and change of surgery from BCT to mastectomy); however, enhancement of benign lesions can lead to unnecessary wider surgery in 3% -6% of patients. 37 At our institution, BCT is the preferred treatment method for managing EBC and selected patients with locally advanced breast cancer. Ultrasonography, mammography and MRI in selected cases are performed at 6 months post-BCT at a time where post-surgical changes may mimic IBTR. We investigated the outcomes and imaging characteristics post-BCT, correlating with histopathological reports.

Method
This was a qualitative, retrospective study of all consecutive adult women (over 18 years) who underwent BCT for EBC and selected cases of locally advanced breast cancer at a specialised breast unit in Inkosi Albert Luthuli Central Hospital over the period 01 January 2011 -31 December 2015 and who were followed up for at least 5 years post-surgery to determine recurrence. All patients must have undergone US, mammography and/or MRI studies during the follow-up period. Patients with benign pathological diagnoses, incomplete histopathological reports, those lost to follow-up or not imaged post-BCT and male patients with breast cancer were excluded.
Clinical, chemotherapy treatment, radiological and histopathological records for patients who underwent BCT for breast cancer during the study period were collected. Relevant patient data collected by means of a data sheet included age, family history, history of previous BCT, comorbidities, clinical assessment (tumour size) and radiological features of the primary tumour as well as the recurrent tumour (size, quadrant, calcifications, mass, distortion, skin thickening, presence of axillary lymph nodes, contralateral breast lesion, distant lesion and enhancement) with respect to US, mammography and/or MRI. Data on patient's radiotherapy treatment were also collected. Histopathological reports of patients who had imaging features which were suspicious for recurrence and had biopsies, were studied. The time, location, imaging features and management of recurrences were collected.
The datasheet consisting of a set of matrices was completed. Each patient was assigned a number on their respective datasheet to ensure patient confidentiality. The radiological images were reviewed and interpreted by two practising radiologists (a qualified specialist and a radiologist in training) and findings recorded on the datasheet. The radiologists were not blinded to one another's assessment. The outcome event (death, disease recurrence, survival) was recorded from the time of surgical intervention to the end of the follow-up period. These data were captured in Microsoft Excel.

Data analysis
Descriptive statistics were used to summarise the data. Frequencies and percentages were used for categorical data. Frequency distributions of numeric variables were examined for normality, means (standard deviation [s.d.]) and median (interquartile range [IQR]). Risk factors associated with recurrence (Yes/No) were examined using Chi-square tests for categorical data and T-test or Mann-Whitney for numeric and ordinal variables. Odds ratios (ORs), 95% confidence intervals (CIs) and p-values were reported. Significant p-value was set at p < 0.05.
Sensitivity and specificity of imaging findings were studied. A multivariable logistic regression was used to identify independent risk factors in patients with recurrence. Data were analysed in Stata ® version 15.1 statistical software.

Ethical considerations
Approval for this study was obtained from the University of KwaZulu-Natal Biomedical Research Ethics Committee with protocol reference number BREC/00002843/2021. All patient identifiers were concealed to ensure patient anonymity.

Patient and disease characteristics
A total of 143 patients underwent BCT during the study period. Fifty-one patients were excluded from the study either because they were lost to follow-up (n = 50) or had benign pathology (n = 1). Therefore, 92 patients who underwent BCT during the study period were included ( Table 1). The mean age was 54.3 years (range: 31-79 years). Most patients were between the ages of 51-79 years (60.9%, 56 patients). The prevalence of HIV in the study population was 8.7% (eight patients) with diabetes mellitus affecting 23.9% (n = 22) of patients. The relationship between comorbid diseases (HIV and diabetes mellitus) was not statistically significant (p = 0.21).

Incidence of recurrence
During the median follow-up period of 70.3 months (range: 1-135.1 months), a total of six patients (6.5%) of the 92 patients who underwent BCT developed IBTR during the study period. The mean time to IBTR was 34.4 months (p = 0.01). The histological subtype of the IBTR was invasive ductal carcinoma in all cases. Four (66.7%) of the six patients with IBTR also presented with distant metastasis. Incidentally, over the study period, one patient presented with a contralateral metachronous DCIS at 67 months and no IBTR, and 11 (12.0%) patients presented with metastatic recurrence only without the presence of IBTR. The frequency of metastatic recurrence sites was bone 56.3%, lung 43.8%, pleura 31.3%, liver 31.3%, axillary lymph nodes 18.8% and brain 12.5%. One patient with IBTR and bone metastasis died during the study period at 13 months. The combined mean time to breast or metastatic recurrence was 32.2 months (p < 0.01).

Risk factors for recurrence
Age < 35 years was significantly associated with IBTR (p < 0.001). Three patients tested positive for a BRCA gene mutation: one patient having BRCA 1 mutation and two demonstrating a BRCA 2 mutation. Two of the patients with BRCA mutation presented with distant metastasis with no evidence of IBTR. The remaining patient had neither IBTR nor distant metastasis. Fifteen patients had a positive family history of breast cancer. Of these, 80% (12/15 patients) had fewer than three first-degree relatives with breast cancer and 20% (three patients) had more than three first-degree relatives with breast cancer. Of the six patients with IBTR, three had a family history of breast cancer: two had fewer than three first-degree relatives with breast cancer and one had more than three firstdegree relatives with breast cancer (p = 0.036).

Pathological factors
The majority of patients had pathological tumour size and extent T2 and pathological lymph node N0 staging correlating to stage II with a mean tumour size of 3.3 cm (range 1.0 cm -5.0 cm) (  other histological subtypes included medullary, papillary, neuroendocrine and undifferentiated cancers. The relationship between IBTR and histological subtype was not statistically significant in this study (p = 0.9).
Estrogen receptor/progesterone receptor (ER/PR) positive Human epidermal growth factor receptor 2 (HER-2) negative was the most commonly expressed hormonal status (61.9%, 57 patients). The risk of IBTR was not associated with hormone receptor positive status (p = 1.00). Grade 1, 2 and 3 tumours represented 9.7%, 53.2% and 23.9%, respectively. Histological tumour grade was not associated with IBTR (p = 0.5). Seventeen (18.5%) patients had lymphovascular invasion with 10 (10.9%) patients demonstrating cancers with perineural invasion. Three of the patients with IBTR also had lymphovascular invasion. The relationship between lymphovascular invasion and recurrence was not scientifically significant (17.6%, p = 0.08). In 63 patients, the lymph nodes were described as clinically suspicious of metastasis; however, only 32 were proven to be pathological on histology. Of these pathological lymph nodes, 10 patients had extranodal extension. Three patients with IBTR had pathological lymph nodes which demonstrated extranodal extension; however, the correlation between IBTR and extranodal extension was not statistically significant (p = 0.42).
Of the 92 patients who underwent BCT, three (4.3%) had involved margins with DCIS after the initial excision. Two of the three patients with involved margins had grade 2 disease and the remaining patient had grade 3 disease. In respect of the management of positive margins, two patients underwent total mastectomy and the remaining patient had a re-excision.
No residual tumour was noted in the re-excision specimen. None of these patients developed IBTR or distant metastasis. A positive margin status did not confer statistically significant results (25%, p = 0.24). A total of 82 patients (89.1%) completed radiotherapy, inclusive of the six patients with IBTR. Reasons for the remaining 10 patients not receiving radiotherapy included: mastectomy for EBC (2/10), overweight for the radiotherapy machine (1/10), development of metastasis before initiating radiotherapy (1/10), no record of radiotherapy (6/10).

Chemoradiotherapy treatment
Of the 82 patients who completed radiotherapy, 44 patients (53.6%) experienced delays in initiating radiotherapy, either due to the compromised radiotherapy service as a result of oncologist shortage during the study period, public worker industrial action, patient confusion with radiotherapy dates or patient financial constraints. Half of the patients with IBTR (3/6 patients) and 47.6% (41/86 patients) without IBTR, experienced delays initiating radiotherapy. However, the correlation between radiotherapy delay and the development of recurrence was not statistically significant (p = 0.9). Five patients (6.1%) experienced interruption in radiotherapy due to skin reactions. None of the patients with IBTR had interruption of radiotherapy. The relationship between interruption in radiotherapy and recurrence was not statistically significant (p = 0.3).

Post-operative imaging findings
Only 80/92 (87.0%) patients were imaged within 1-year post-BCT (Table 3). Of these, only 22 patients (27.5%) were imaged by both US and mammogram. Fifty-two patients (65%) had only a mammogram, four (5%) only US and two (2.5%) only MRI in the first post-BCT imaging series. Based on the post-BCT images, a total of 32 patients (40%) underwent biopsy to confirm radiological suspicion of recurrence. In seven of these patients the biopsy returned a positive diagnosis for malignancy; six IBTR and one contralateral metachronous cancer ( Figure 1).
An asymmetrical density or mass was found in 46.9% (15/32) patients who underwent biopsy (p < 0.001). Of these patients, asymmetric density was found in four patients with IBTR and 11 patients without IBTR with an OR of 4 (95% CI: 0.38-41.74, p = 0.25). Increasing breast oedema was observed in 9.4% (3/32) patients who underwent biopsy (p = 0.03). None of the patients who had post-operative increasing breast oedema had an IBTR (p = 0.9). Calcifications were reported in nine patients (28.1%) who underwent biopsy: coarse (four), pleomorphic (three), indistinct (two) (Figure 2 and Figure 3). All patients with pleomorphic calcifications had biopsyproven recurrence. None of the patients with coarse and indistinct calcifications developed recurrence. The presence of calcifications was associated with IBTR with OR of 4 (95% CI: 0.53-30.16, p = 0.3). Skin thickening (> 3 mm) was described in three patients who were biopsied. Recurrence occurred in one patient who had skin thickening (> 3 mm) with OR of 2.5 (95% CI: 0.18-34.67, p = 0.47). This patient also had a breast mass on mammogram in addition to skin thickening. Of the 25 histologically negative biopsies, fat necrosis accounted for 18 while four represented normal fibrofatty breast tissue, one fibroadenoma, and one each for fibrous tissue and blood ( Figure 4 and Figure 5). Benign postoperative changes were indistinct or coarse calcifications, absence of lymphadenopathy and absence of skin thickening (> 3 mm).
A comparison between mammographic findings of IBTR and the histopathological reports was made. In this cohort, we found that the sensitivity and specificity of predicting breast cancer recurrence on mammography were 100% and 70.6% respectively with a positive and negative predictive value of 21.9% and 100%, respectively.
MRI was utilised to complement US or mammogram in six additional patients due to dense breasts. Collectively, MRI was used in eight patients during the first round of postoperative imaging which resulted in two positive diagnoses of breast recurrence ( Figure 6).

Discussion
This study was performed to review outcomes following BCT in a single tertiary centre. Nearly all the tumours were pT1 and pT2 lesions in keeping with international standards and guidelines for BCT in EBC. 3,38 This study determined a relative IBTR of 6.5% with a mean time to recurrence time of 34.4 months and is comparable to local and international recurrence rates which are reported as acceptable if between the 5% -15% range. 1,11,14,16,19 Age < 35 years was strongly associated with IBTR in our cohort, in keeping with published literature. The margin positivity was 3.3%; however, none of these patients with positive margins developed IBTR. Although margin status is associated with a higher IBTR, we could not demonstrate this relationship in the current study possibly due to early reexcision which mitigated the recurrence risk and the small number of patients in our series.
Absence of pathological complete response to neoadjuvant chemotherapy is associated with increased IBTR; however, in  this study this correlation was not significant. 27 Radiotherapy is known to be protective against IBTR. 39,40,41,42,43 However, only 82/92 (89.1%) patients completed radiotherapy in this study. Contrary to current literature, we could not demonstrate the relationship between omission of radiotherapy and recurrence (p = 0.9). In the current study, the correlation between interruption and delays in radiotherapy and IBTR was not statistically significant (p = 0. recurrent malignancy. 5,6 Imaging features that were associated with recurrence in our series are similar to those reported in the literature, and included an enlarging mass, microcalcifications, increased skin thickening (> 3 mm), increasing oedema and asymmetric density. 5,6 In the current study, asymmetric density and microcalcifications most commonly correlated with recurrence, followed by skin thickening. The lack of statistical significance may be due to the small sample size. Contrary to current literature, we could not demonstrate a relationship between linear, coarse and indistinct calcifications and recurrence. In this study, only pleomorphic calcifications were seen with recurrence.
The sensitivity and specificity of predicting breast cancer recurrence on mammography are 100% and 70.6% respectively with a positive and negative predictive value of 21.9% and 100%, respectively. This sensitivity is comparable and slightly better than international standards which describe a sensitivity on mammography of 55% -68%. 5,44,45 We anticipate that the use of contrast enhanced mammogram at our institution will assist in improving diagnostic imaging accuracy. 46 Features which correlated with benign breast changes included absence of skin thickening (> 3 mm), indistinct and coarse calcifications, absence of lymphadenopathy.
The study also revealed a lack of adherence to a standard post-treatment surveillance protocol. Of the 80 patients imaged within 1-year post-BCT, only 22 (27.5%) were imaged by both US and mammogram. Factors influencing poor adherence to imaging protocol include patient factors (financial constraints) and non-functional mammogram machines at the base hospitals. It is therefore imperative to ensure adequate access to imaging during the surveillance period to adequately assess the breast.

Limitations
This study has a number of limitations due to the retrospective nature of the study. It was a single centre study, where most patients presented with locally advanced disease; therefore, the number of patients suitable for BCT was small. This number was further reduced by patients lost to follow-up during transfer to off-site oncology centres. These factors contributed to the small sample size.
There was limited access to the adjuvant treatments as patient files were transferred to the off-site oncology follow-up facility site. Where files were available, some information was incomplete due to poor record keeping and nonstandardisation of patient records which is a general challenge in the public sector.

Conclusion
The study demonstrates that BCT outcomes at our centre are within local and international standards in terms of margin positivity and recurrence rate. The accuracy of assessing the post-BCT breast for recurrences is in keeping with international standards. The sensitivity and specificity of diagnosing recurrence on mammogram surpasses current international standards. Although the correlation between delay and interruption of radiotherapy to IBTR was not statistically significant in this study, it has highlighted patient and systemic barriers to healthcare. Implementation of efficient decentralised breast imaging centres could improve patient adherence and timeous annual breast imaging following breast cancer therapy.