Discordance between Primary Breast Cancer and Ipsilateral Breast Cancer Tumor Recurrence as a Function of Distance

Background: Risk factors for ipsilateral breast cancer tumor recurrence (IBTR) are well established and include grading, nodal status, and receptor status. Little is known about the influence of the local distance between the primary tumor and recurrences on changes in tumor characteristics and prognosis. Methods: In a retrospective setting, we analyzed primary breast cancers and their recurrences. Localizations of primary and recurrent breast cancer were recorded to calculate the relative distance in pixels. Analysis was performed regarding tumor characteristics, relative distance between both, and their impact on breast cancer prognosis. Results: In a cohort of 142 patients with ipsilateral recurrence, no statistically significant difference could be shown in the change in tumor characteristics depending on distance. Progesterone receptor (PR) and estrogene receptor (ER) status changed in 22.7% and 14.9% of cases, respectively. human epidermal growth factor receptor 2 (ERBB2, HER2) status changed in 18.3% of cases. Survival was in accordance with the literature, with luminal-A-like tumors as best and triple negative breast cancers (TNBC) as worst prognosis. With a threshold of 162 pixels, the survival was significantly better in the group with shorter distance. Conclusion: Change in tumor characteristics from primary breast cancer to recurrence occurs more often in PR than ER. In contrast to other work, in this dataset, recurrences with a larger distance to the primary tumor had a worse prognosis in univariate analysis. A Cox model might indicate the possibility that this influence is independent of other risk factors.


Patients
This is a retrospective study including patients who were newly diagnosed with breast cancer at the University Breast Center of Franconia, Germany, between 1995 and 2010. All breast cancer patients were screened, and patients with ipsilateral breast cancer tumor recurrence were identified (n = 364). For this study, patients were excluded for the following reasons: missing data on immunohistochemistry at primary diagnosis or recurrence, missing imaging data, or patients who did not have salvage surgery.
Finally, a complete dataset of 142 patients with ipsilateral breast cancer recurrence was obtained, with ER and/or PR status available from both primary diagnosis and recurrence. Additionally, both Ki-67 and HER2 status was reported at primary diagnosis and recurrence, wherever available. The study was in accordance with the Declaration of Helsinki and local ethical guidelines.

Data Collection
The University Breast Center of Franconia received certification from the German Cancer Society (Deutsche Krebsgesellschaft) and the German Society for the Study of Breast Diseases (Deutsche Gesellschaft für Senologie) in 2004. To obtain certification, a breast center has to document each case of breast cancer, including patient and tumor characteristics, treatment data, and some epidemiological data into a prospective database. Follow-up information regarding local recurrences, distant metastases, and death has to be provided for up to 10 years after the initial diagnosis. Follow-up was done according to local guidelines with mammography and breast ultrasound every 6 to 12 months. All histopathological data also have to be documented from the original pathological reports, including tumor size, axillary lymph node status, grading, and ER, PR, and HER2 status. Breast centers and their data quality are audited annually as part of the continuous certification process. Data obtained through these processes were used in the analysis presented here.

Histopathological Assessment
All of the histopathological information used in this analysis was directly documented from the original pathology reports, which were reviewed by two investigators. Grading, tumor type, ER status, PR status, HER2 status, and proliferation status (as assessed by Ki-67 staining) have been routinely recorded at the breast center since 1995, and were performed on formalin-fixed, paraffin-embedded tumor tissue. Monoclonal mouse antibodies against estrogen receptor-alpha (clone 1D5; 1:200 dilution, DAKO, Glostrup, Denmark), monoclonal mouse antibodies against the progesterone receptor (clone pgR636, 1:200 dilution, DAKO, Glostrup, Denmark), and monoclonal mouse antibodies against Ki-67 (clone MIB-1, 1:200 dilution, DAKO, Glostrup, Denmark) were used to stain the primary tumors and recurrences. The percentage of positively stained cells was included in the pathology reports. The tumors were considered to be positive for the estrogen and progesterone receptors if 1% or more of the cells showed positive staining.
A polyclonal antibody against HER2/neu (A0485, 1:200 dilution, DAKO, Glostrup, Denmark) was used, and HER2 status was stated in the pathology reports as negative, 0+, 1+, 2+, or 3+. As for a large proportion of recurrences, FISH or CISH testing was not available, so the use of those data was not taken into consideration to further adjust the HER2 status. Scoring was carried out in a standardized way by a group of dedicated pathologists in routine surgical pathology. With regard to Ki-67, areas with the highest Ki-67 labeling were investigated, and approximately 500 cells were counted with 400-fold magnification.

Documentation of Localization
The positions of the primary breast cancers and the recurrences were located with a mammogram, breast ultrasound, and palpation, and were documented in the patient chart.
Using an Intuos3 professional pen tablet (Wacom Europe Ltd., Krefeld, Germany) and the ImageJ picture editing program (Research Services Branch, National Institute of Mental Health, Bethesda, MA, USA), a standard picture of the upper female body was adjusted, and the coordinates of the localization of the primary tumor were marked and noted. The coordinates of the recurrence site were then marked and noted. This resulted in two coordinates (x, y) of each site. To ensure consistency in the digitizing procedure, the position of the pictograms on the pen tablet was standardized ( Figure 1).

Statistical Considerations
Overall survival (OS) was defined as the time from the date of primary breast cancer diagnosis to either the date of death or the date of censoring. Additionally, we examined survival after recurrence when local distance between primary diagnosis and recurrence was considered. Patients who were lost to follow-up within 20 years were censored at the last date they were known to be alive. For example, a patient who was alive 20 years after diagnosis would have been censored at that point.

Statistical Considerations
Overall survival (OS) was defined as the time from the date of primary breast cancer diagnosis to either the date of death or the date of censoring. Additionally, we examined survival after recurrence when local distance between primary diagnosis and recurrence was considered. Patients who were lost to follow-up within 20 years were censored at the last date they were known to be alive. For example, a patient who was alive 20 years after diagnosis would have been censored at that point. Survival rates were estimated by the Kaplan-Meier product limit method and illustrated by Kaplan-Meier curves. Differences of survival between groups were tested with log-rank tests. A Cox regression model was used to examine the relationship between survival and the local distance correcting for the influence of several tumor characteristics, age and time to recurrence. To identify the most appropriate model, stepwise variable selection was performed.
Group comparisons unrelated to survival were done with the Kruskal-Wallis test and the Mann-Whitney U test, respectively. To account for multiple testing, Benjamini-Hochberg correction was performed where appropriate.
All of the tests were two-sided, and a p value of ≤0.05 was regarded as statistically significant. Calculations were carried out using the SPSS software package (SPSS Statistics for Windows, version 24, IBM Corporation, Armonk, New York, NY, USA) and the statistical programming language R V3.6.3 [26].

Patient Characteristics
The final cohort consisted of 142 patients. The average age at primary diagnosis was 55.01 years. The average age at menarche was 13.66 years. Overall, 106 stated that they had never used hormone replacement therapy (HRT), and 22 had used HRT at some point. All patients were treated according to the German guidelines for endocrine therapy, chemotherapy, trastuzumab, or radiation therapy depending on the subtype of breast cancer recurrence. Only patients with negative resection margins after the surgery for primary breast cancer were included in this analysis. The median follow-up time was 158 months.
Further patient characteristics can be seen in Figure 2 in the Kaplan-Meier curve for mean survival (months) by distance (Table 1). A Cox regression model was used to examine the relationship between survival and the local distance correcting for the influence of several tumor characteristics, age and time to recurrence. To identify the most appropriate model, stepwise variable selection was performed.
Group comparisons unrelated to survival were done with the Kruskal-Wallis test and the Mann-Whitney U test, respectively. To account for multiple testing, Benjamini-Hochberg correction was performed where appropriate.
All of the tests were two-sided, and a p value of ≤0.05 was regarded as statistically significant. Calculations were carried out using the SPSS software package (SPSS Statistics for Windows, version 24, IBM Corporation, Armonk, New York, USA) and the statistical programming language R V3.6.3 [26].

Patient Characteristics
The final cohort consisted of 142 patients. The average age at primary diagnosis was 55.01 years. The average age at menarche was 13.66 years. Overall, 106 stated that they had never used hormone replacement therapy (HRT), and 22 had used HRT at some point. All patients were treated according to the German guidelines for endocrine therapy, chemotherapy, trastuzumab, or radiation therapy depending on the subtype of breast cancer recurrence. Only patients with negative resection margins after the surgery for primary breast cancer were included in this analysis. The median follow-up time was 158 months.
Further patient characteristics can be seen in Figure 3 in the Kaplan-Meier curve for mean survival (months) by distance (Table 1).

Tumor Characteristics
In 44.4% of cases, the recurrence was located in another quadrant to the primary tumor of the same breast. In 55.6%, the recurrence was in the same quadrant as the primary carcinoma.
Overall, 54.9% of the primary tumors were located in the upper outer quadrant, 23.9% were in the upper inner quadrant, and 7.7% and 13.4% were in the lower outer and lower inner quadrants, respectively.
In addition, 8.5% of the tumors at primary diagnosis were well differentiated, 54.2% were moderately differentiated, and 32.4% were poorly differentiated ( Table 2).
Tumors were categorized in four subtypes by hormone receptor status, grading, HER2-status and proliferation rate   Regarding the tumor subtype at recurrence, the mean distance between the site of primary diagnosis and site of recurrence (in pixels) was 86 for luminal-A-like recurrences, 144 for luminal-B-like recurrences, 201 for HER2-enriched recurrences, and 138 for TNBC. These distances did not differ significantly (Kruskal-Wallis test: p = 0.067) ( Table 4).

Grading and Change in Grading between Primary Diagnosis and IBTR with Regard to Distance
The mean local distance between primary site and recurrence for G1, G2, or G3 primary tumors corresponded to 101 pixels, 133 pixels, and 149 pixels, respectively. Regarding the grading at recurrence, the distances were 99 pixels, 137 pixels, and 134 pixels, respectively. The distances did not differ significantly between groups with different changing behavior of grading from the primary site to recurrence: (a) primary G1/2 →≥ recurrenceG1/2; (b) primary G1/2 → recurrence G3; (c) primary G3 → recurrence G1/2; (d) primary G3 → recurrence G3 (Kruskal-Wallis test: p = 0.912).

Change of Tumor Characteristics
The ER status changed from positive to negative in 9.2% of the cases, and from negative to positive in 5.7% of the cases, between primary breast cancer diagnosis and breast cancer recurrence.
Regarding the four possible groups (negative to negative, negative to positive, positive to negative, and positive to positive), the mean distances were 148, 107, 131, and 131 pixels, respectively. No significant difference could be shown (Kruskal-Wallis test: p = 0.789).
A change in the PR status from positive to negative occurred in 14.2% of cases, and from negative to positive in 8.5% of cases. In addition, no significance could be shown in the distances (p = 0.103).
HER2 status changed from positive to negative in 8.1% of cases, and from negative to positive in 10.2% of cases. There was no significance regarding distance (p = 0.652).

Distance between Primary Site and Recurrence Depending on Nodal Status and Tumor Size
The mean distance between primary diagnosis and IBTR was 111 pixels for nodal negative patients, and 166 pixels for nodal positive patients. The Mann-Whitney U-test revealed a p-value of 0.034, but after Benjamini-Hochberg correction for multiple testing, no significant difference could be shown (p = 0.084 after correction).
The mean distance for smaller tumors (T0-2) was 128 pixels, whereas the mean distance for larger tumors (T3+4) was 215 pixels. A significant difference can be seen between these two groups (p = 0.005; Benjamini-Hochberg corrected p = 0.038)

Age, BMI, and Time to Recurrence in Relation to Distance of Recurrence
No significant relationships between distance and age, BMI, or time to recurrence were observed by means of correlation analysis after a Spearman test (p = 0.512; p = 0.772; p = 0.592).

Survival by Tumor Subtype at Primary Diagnosis
The mean survival rate was highest within the luminal-A-like primary tumor group (190.5 months), followed by the luminal-B-like group (154.8 months), then the HER2-enriched group (142.0 months), and finally the TNBC group (118.2 months). This was significant in the log rank test (p = 0.043) (Figure 3).

Survival as a Function of Local Distance
We determined a threshold for local distance by creating two groups using maximally selected rank statistics determined by log-rank tests. The optimal threshold value was found to be 161.56 pixels. The five-year survival rate for the first group (≤162 pixels) was 70%, whereas the second group (>162 pixels) was 38%. The Monte Carlo simulated global p value for the maxstat test was p = 0.043. The Kaplan-Meier curves are presented in Figure 3.
For the cox regression, stepwise variable selection was performed and the categorical distance group determined in the previous analysis (≤162 vs. >162 pixels), PR, tumor size, and time to recurrence were selected. Distance was a significant factor with HR = 1.83 and a p-value of 0.034, as were a positive PR status (HR = 0.47, p = 0.007), tumor size T3/4 (HR = 3.804, p < 0.001), and time to recurrence in years (HR = 0.867, p = 0.021) ( Table 5). In a similar model with distance in pixels as a metric variable instead of a categorical distance grouping factor, distance was not significant.  We determined a threshold for local distance by creating two groups using maximally selected rank statistics determined by log-rank tests. The optimal threshold value was found to be 161.56 pixels. The five-year survival rate for the first group (≤162 pixels) was 70%, whereas the second group (>162 pixels) was 38%. The Monte Carlo simulated global p value for the maxstat test was p = 0.043.

Discussion
In this study, 142 patients with breast cancer were reviewed, who subsequently developed IBRT. By analyzing primary tumor and recurrent tumor characteristics and local distance between primary and recurrent breast cancer, the objective was to find correlation between changes in tumor characteristics as a function of local distance.
As with metastases, biomarker characteristics in IBTR change over time. PR changes at a higher frequency than ER, both from negative to positive and vice versa. These results are similar to those from other studies. The findings regarding time to recurrence by subtype and survival by subtype (with luminal-A-like tumors having the best prognosis and TNBC having the worst prognosis) are consistent with the literature and are not surprising [3,[14][15][16].
In addition, the analysis of survival by nodal status and tumor size are in accordance with other work [3,[22][23][24][25], and reflects clinical experience.
Recurrences of smaller tumors (T0-2) occur significantly closer to the primary site than those of larger tumors. A possible explanation could be the area covered by boost radiation. Unfortunately, we have no data about the radiation fields, therefore we could not further evaluate this.
Interestingly, no significant difference in local distance between the primary site and recurrence could be shown in the different subtypes, either by subtype of primary tumor or of recurrence. Regarding the Cox model, a positive PR-status (HR = 0.470), larger tumors (HR = 3.804) and time to recurrence (HR = 0.867) are in accordance with known risk factors and are not surprising [20]. In our Cox model, local distance is a significant factor in survival (HR = 1.825). As there is no significant change in biomarkers regarding distance, the distance itself seems to play an important role. Most likely, a substantial number of recurrences occurred in the axillary pit (Figure 1b,e) and were characterized as IBTR following the given definition. It is known that recurrences in the axillary lymph nodes have a worse prognosis than recurrences in the breast [20]. Moreover, the distance to a recurrence in the axillary pit is often larger than the distance to a recurrence in the breast. This might be an explanation as to why a larger distance between the primary tumor and IBTR has a worse prognosis. However, as the cutpoint was selected by optimization, the Cox model may not be entirely accurate. It should therefore only serve for generating hypotheses, and should be further evaluated in future studies.
In the Kaplan-Meier analysis, on the other hand, the reported significant p-value regarding survival differences in groups defined by the optimal threshold for distance takes this optimization into account and is valid. This is one of the first descriptions of survival relating to the distance between the primary tumor site and the recurrence site. The aim of this study was to examine the change in receptor status between the primary tumor and IBTR. The changes in ER and PR were in accordance with the literature and clinical experience. No significant differences were seen with regard to local distance for ER, PR, or HER2 status.
One study [3] showed a significantly better prognosis in the group in which the recurrence occurred elsewhere in the breast compared to that in which the recurrence occurred at the same site.
Another study evaluating the location did not find any significant differences if the IBTR was within 3 cm of the primary tumor or not [20].
Our results regarding survival and local distance are in contrast to those results. However, in the first study, the patients were all treated before 1984, and in 64% of all cases, the ER status was unknown; furthermore, the treatment of breast cancer in general changed, so these results cannot really be compared. In the second study, the location was determined as the relative position to the index tumor by clinical information or mammographic reports. In contrast to our study, where the location was determined by mammograms and ultrasound pictures, the evaluation might be more precise.
One limitation of our study is the rather small sample size, but a long follow-up of 158 months makes the dataset valuable.
The lack of an absolute distance is a problem, as we only have a relative distance on a standardized breast. Therefore, the exact breast size and distance was not taken into account. The relative size of the breast in the pictogram can serve as a reference. For example, the total width of the upper body in the pictogram was 960 pixels, so 160 pixels corresponds to about 1/6 of the total width.
Altogether, the changes in biomarkers between the primary tumor and IBTR are relevant for clinical practice, and therefore the immunohistochemical profile has to be evaluated in cases of breast cancer recurrence. Not only should the changes in biomarkers influence the treatment of IBTR, but also the local distance of the recurrence should be evaluated, to clarify the overall risk to the patient.

Conflicts of Interest:
The authors declare no conflict of interest.