CGH has greatly facilitated the detection of gains and losses in DNA copy number, and is especially useful when elucidating patterns of genomic alterations in tumors. Since the extent of genomic imbalances can be determined with greater mapping precision due to the high resolution and large number of data points , aCGH in particular may be of clinical value in differentiating new primary tumors from recurrent lesions and genomic analysis can help reveal the relationship between multiple tumors. Moreover, this may have significant implications for selection of optimal adjuvant treatment. Today, differences in tumor histopathology are used to distinguish between disseminated disease and the occurrence of multiple synchronous primary tumors. Recent reports, however, demonstrate that histopathological evaluation fails to provide unambiguous evidence for tumor origin; hence it has been proposed that genetic analyses are more reliable in this context [9, 14, 16]. In the present study we elucidated genomic alterations in synchronously diagnosed unilateral and bilateral breast tumor pairs using 32 K tiling BAC aCGH. The analysis of tumor pairs from 10 patients diagnosed with synchronous unilateral breast cancer by unsupervised hierarchical clustering resulted in tumors from five patients clustering together as pairs, suggesting a genomic similarity and a possible common origin of the two tumors in these patients. These tumor pairs clustered together despite complex chromosomal copy number aberrations. In line with our findings, previous studies based on genetic as well as histopathological evaluations have concluded that a large fraction of unilateral breast tumors arise through intramammary dissemination of a single breast cancer, i.e. suggesting that most unilateral tumors develop from the same clonal origin [11, 16, 34–36]. More recently, similar results have been obtained by analyzing unilateral breast tumors using metaphase CGH and unsupervised hierarchical clustering; one such study revealed that paired tumors from three of four unilateral breast cancer patients demonstrated a clonal relatedness based on copy number profiles . Alternatively, it is possible that the genomic similarities observed in our study could be caused by a 'field effect' whereby e.g. a common environmental exposure could cause the development of genomically similar, but independent, tumors. However, this appears less likely, as such an environmental effect would be expected to affect the similarity also among bilateral cases, which was not the case. In our study, an absence of genomic similarities was found in five of ten synchronous unilateral tumor pairs, i.e. the two tumors from the same patient did not cluster together. Two of these tumor pairs displayed highly discordant chromosomal imbalances, emphasizing the possibility that the tumors have developed independently. Alternatively, intra-tumor heterogeneity may explain these differences, as recently suggested . Based on these findings we conclude that synchronous unilateral tumors often are genetically similar, indicating that they develop from a single clonal origin, but multiple individual primary carcinomas might also develop synchronously in the same breast, confirming earlier studies [8, 16, 19, 36].
Only one of eight synchronously diagnosed bilateral breast tumor pairs displayed genomic similarities in the clustering analysis. Based on clinical as well as histopathological variables, this patient most likely presented with generalized disease, further underscoring the clonal relatedness between the tumors. This patient also received radiation and chemotherapy prior to surgery, which may potentially have affected the genomic analyses; however this does not seem to be the case because no pathological response to the treatment was observed. Histopathological parameters that support the notion of contralateral spread include positive lymph nodes on both sides, the same histological grade (3) and type (ductal carcinoma), all of which have been suggested as evidence for metastatic disease [1, 3]. In contrast, both tumors displayed intraductal carcinoma, one of the criteria used to determine lesions of independent origin in bilateral breast cancer .
Earlier studies of bilateral breast cancer have demonstrated a failure to provide unambiguous evidence for the distinction between independent tumors and contralateral spread based on histopathological parameters [10, 14, 15]. Of the tumor biological and histopathological parameters investigated in our study, DNA index and histological type correlated with the genetically based clustering outcome. Specifically, all six tumor pairs that displayed genomic similarities within the pair also had highly similar DNA indices, and both tumors within the pair were of the same histological type in these cases. Among the tumors that did not illustrate pair-wise genomic similarities in the clustering analysis, two tumor pairs had different histological types (tumors 1b and 4a were of lobular type and tumors 1a and 4b were of ductal type), and seven of ten tumor pairs displayed large differences in DNA index within the tumor pairs. Tumor size, hormone receptor status, distance between tumors (for unilateral tumors) and lymph node status were inconclusive or did not correlate with the results of the genomic analysis. Nevertheless, one study has demonstrated that the distance between the tumors might be of significance when discriminating between independent tumors and intramammary dissemination in multi-focal or multi-centric unilateral tumors; the study was however very small and no significance was reached .
Since tumor histology (type and grade) is inconclusive in most cases of multiple tumors in the same breast, it is difficult to discriminate between the development of multiple primary lesions and lesions developed from the same tumor origin. Based on the notion that intraductal cancer constitutes a pre-invasive phase in breast cancer progression, it has been suggested that the presence of intraductal components in each of the lesions indicates independent origins [7, 8]. However, more recent studies suggest that the occurrence of an intraductal component is of little help in distinguishing multiple independent tumors from intramammary dissemination [16, 35, 36]. The fact that intraductal carcinoma occurred at the same frequency in clustered pairs as in non-clustered pairs in our study (four of five vs. three of five), together with the observation that the grade of the intraductal component was similar to that of the corresponding invasive component, suggests that the intraductal component represents an intraductal growth phase rather than pre-invasive in situ cancer and confirms the view put forward in the latter studies. By histopathological examination, it is very difficult to distinguish between pre-invasive carcinoma in situ and intraductal spreading of invasive components. However, in our study, the unilateral tumor pairs that clustered together (four of five) demonstrated vessel invasion while none of the pairs that did not cluster displayed this feature, suggesting a common origin of the tumors that clustered together. This stresses the importance of determining the ability of a tumor to spread and invade surrounding tissue by distinguishing between a pre-invasive in situ component and intraductal spreading by evaluating vessel invasion in conjunction with the intraductal component.
Even if a majority of unilateral breast tumors arise through intramammary dissemination of a single breast cancer a significant number of synchronous unilateral tumors could, according to our study, have developed from independent origins. This might be important to consider when evaluating the clinical outcome of the patient. Our findings demonstrate the importance of evaluating the properties of both tumors in order to determine the most optimal treatment for the patient. The possibility of synchronous unilateral tumors representing individual clones could have implications for the clinical outcome of the patient if both tumors are not accurately characterized. Using a PCR-based approach, it was recently shown that clonally related ipsilateral recurrences were more frequently of higher histological grade and developed sooner after initial treatment than clonally distinct ipsilateral recurrences , and it has also been reported that women with clonally related ipsilateral recurrences had poorer outcomes than patients with unrelated tumor pairs . In the present study, all the unilateral tumor pairs that clustered together were histological grade 3, while the non-clustering tumor pairs were low to moderate histological grade (see Table 1). Molecular approaches may hence provide a reliable means of identifying patients who may benefit from more aggressive systemic treatment, i.e. whose tumors are clonally related and more likely to be associated with metastatic disease. Of interest, a recent study of 22 pairs of ipsilateral breast cancers revealed a statistical difference in metastasis-free survival between new primary tumors and recurrences as defined by a 'partial identity score' based on DNA breakpoint information, suggesting that genomic analyses could outperform clinical and histopathological characteristics in terms of prediction of prognosis , and supporting our conclusions. The importance of correctly characterizing the individual synchronous tumors is also apparent in bilateral breast cancer. Even if rare, bilateral metastatic spread seems to occur, and since metastatic disease is known to result in a worse prognosis it is important to consider the occurrence of metastatic disease in these cases by analyzing both tumors. Our results clearly demonstrate that through the evaluation of both tumors in the pair by genome-wide aCGH and hierarchical clustering analysis one might obtain objective information regarding whether the tumors are independent lesions or disseminated disease.