Receptor conversion in distant breast cancer metastases

With 1,000,000 new cases causing 375,000 deaths worldwide per year, breast cancer is the leading cause of female cancer death worldwide [1]. Early detection, optimal surgery and adjuvant therapy are the key strategies to improving prognosis. Nevertheless, about one third of patients will develop distant metastases and eventually die of the disease. Patients who develop distant metastases usually undergo systemic therapy with chemotherapy, hormonal therapy and/or human epidermal growth factor receptor 2 (HER2) targeted therapy. Choice of therapy is currently personalized on the basis of the immunophenotype of the primary tumor, since distant metastases are often not biopsied, partly because of limited accessibility of these metastases, but also because it is not considered necessary for further therapeutic decision making.

However, previous studies [2‐23] have indicated that receptor status of breast cancer metastases may differ from the primary tumor, generally denoted "receptor conversion". The published studies suggest that, compared to the primary tumors, estrogen receptor (ERα) and progesterone receptor (PR) are more frequently negative in distant metastases, whereas HER2 is more often positive. These observations, if confirmed, have important clinical consequences, since this would mean that a number of patients are withheld adequate systemic treatment for their metastases. In addition, if immunophenotype conversion would occur in high frequency, this would make it clinically very relevant to biopsy (even difficult to access) distant metastases to assess hormone receptor and HER2 status. An alternative for taking biopsies could be molecular imaging methods like positron emission tomography (PET) and single photon emission computed tomography (SPECT) that are currently being developed to functionally assess immunophenotype of breast cancer metastases [24‐26].

Unfortunately, the previous conversion studies suffer from several limitations: small size (six studies contain <30 cases and eight studies ≤62 cases) [2, 3, 6, 7, 9, 10, 12‐15, 18‐20, 22], only one metastatic site studied [9, 18], using a ligand-binding assay [2, 3, 7, 10], which, especially in the case of metastases, may be biased by low cellularity and contamination by nonmalignant cells, inclusion of decalcified bone metastases [2, 4, 6, 8‐10, 12, 16, 19, 20, 22] that may give rise to false negative immunohistochemistry, extraction of original immunohistochemistry results from the pathology report instead of repeating the staining [3, 5, 8, 10, 11, 23], and/or use of tissue arrays [21] which may introduce sampling bias. Therefore, the available data may not be sufficiently reliable to change current clinical practice, although several guidelines already advise to rebiopsy distant metastases when possible [27, 28]. Consequently, for most patients with metastatic breast cancer hormone receptor and HER2 status in the primary tumor are still used to guide therapy.

We have now performed a large study analyzing metastases from different sites, while restaining metastases and primary tumors side-by-side with optimal current immunohistochemical methods for ERα, PR and HER2 on full sections to asses the conversion rate of ERα, PR and HER2 status in distant metastases compared to the primary breast carcinomas.

The percentage of nuclei expressing ERα or PR was generally lower in the distant metastases than in the primary tumor (Figure 1), but significance was only reached for PR (P <0.001). Receptor conversion exceeding the threshold of 10% occurred for ERα in 10.3% and for PR in 30.0% of the patients (Table 2). Such conversion was mainly from positive to negative: 10.7% of the patients converted from ER+ or PR+ to ER-/PR-, and 3.4% from ER-/PR- to ER+ or PR+. Receptor conversion exceeding the threshold of 1% occurred for ERα in 15.1% and for PR in 32.6% of the patients (Table 2), while 12.4% of the patients converted from ER+ or PR+ to ER-/PR-, and 8.2% from ER-/PR- to ER+ or PR+.

Receptor conversion for ERα and PR (10% threshold) seemed to occur especially in liver (ERα 12.7%, PR 41.2%) and brain metastases (ERα 13.7%, PR 36.3%) (Table 3). For PR, conversion was significantly more often seen for brain, liver and gastro-intestinal metastases (P = 0.04).

For HER2, receptor conversion by IHC occurred in 5.2% of patients, about half of them from negative to positive and the other half from positive to negative (Table 2). Receptor conversion for HER2 seemed to occur especially, although not significantly, in liver metastases (9.5%) (Table 3). Of the 12 cases that showed HER2 conversion by IHC, 5 showed also conversion by SISH. One further case showed conversion by ISH, but not by IHC (Table 4).

Examples of conversion of ERα, PR and HER2 from primary breast cancers to distant metastases are shown in Figure 2.

Previous studies have shown that distant breast cancer metastases may show receptor conversion, potentially leading to inappropriate choices of systemic treatment in these patients. These previous studies however suffered from several limitations not allowing the researchers to draw definite conclusions that may change clinical practice. We therefore set out to reevaluate receptor conversion in a large group of non-bone distant breast cancer metastases using optimal methodology. Receptor conversion by IHC for ERα, PR and HER2 occurred in 10.3%, 30.0% and 5.2% of patients, respectively, using the traditional 10% threshold. When using the new 1% ERα and PR threshold according to the ASCO guidelines, conversion rates were even higher at 15.1% and 32.6%.

Previous studies (all using the 10% threshold when indicated) reported ERα receptor conversion rates from 12% to 54%, clearly higher than in the present study. The explanation for this finding may be that previous ERα studies did not restain both the primary and metastatic lesions [3, 5, 8, 10, 11], used ligand-binding assays [2, 3, 7, 10] (where the result is influenced by differences in the percentage of non-tumor cells in the samples), or included bone metastases that may suffer from false negative IHC results due to decalcification [2, 4, 6, 8‐10, 12]. In addition, with 233 cases our series is much larger than most previous ERα studies. Only two previous studies report on comparable numbers of cases (200 and 211, respectively), but in these studies the original immunohistochemistry results from the pathology report were used instead of renewed stainings. Previous studies (all using the 10% threshold when indicated) reported PR conversion rates from 28% to 61%, again higher than in the present study. The explanation for this may be similar to what has been mentioned above: previous studies did not restain both the primary and metastatic lesions [5, 8, 10, 11], used ligand-binding assays [2, 7, 10], or included bone metastases [2, 6, 8‐10, 12]. Further, our series is much larger with 233 compared to most previous PR studies with 9 to 59 cases. There were two previous studies with respectively 173 and 211 cases, but these used the original immunohistochemistry results from the pathology report instead of renewed stainings.

Conversion for ERα and PR was mainly from positive in the primary tumor to negative in the metastases as has been described before [2, 4, 6, 8, 10‐12]. This finding may well be explained by clonal selection of less differentiated receptor negative cells during the metastatic process, for example, elicited by adjuvant hormonal treatment [12, 35, 36]. However, in a few cases conversion from negative in the primary tumor to positive in the metastases occurred. This phenomenon has also been described before [2, 3, 8, 10, 11], but is more difficult to explain. Although false negative primary tumor results cannot be fully excluded, we also assessed the adequacy of staining by analysis of staining of epithelial cells in the normal ducts and lobules as an internal control. Perhaps in these cases small receptor positive clones within the primary tumor preferentially metastasized [37]. Alternatively, this phenomenon could be a result of genetic drift during tumor progression [38]. Previous studies reported HER2 immunophenotype conversion rates from 0% to 58.3%, which is generally higher than the 5.2% conversion rate we have found in the present study. As mentioned above, previous HER2 studies did not restain both the primary and metastatic lesions [10, 11, 23], used TMAs [21] or included bone metastases [10, 12, 16, 19, 20, 22]. Further, our series is much larger with 233 cases compared to most previous studies with 12 to 211 cases, and in the only larger study (N = 382) restaining was not performed. By SISH, only half of these IHC conversions were accompanied by a difference in HER2 gene amplification status. Therefore, for only 6 out of the 233 patients (3%) a "true" conversion on the gene level of HER2 status between primary tumor and distant metastasis could be demonstrated. The fact that scoring of HER2 SISH signals is more straightforward than interpretation of HER2 IHC may play a role here. Contrary to ERα and PR that preferentially converted from positive in the primary tumors to negative in the metastases, HER2 receptor conversion occurred both ways. Although from a tumor progression model one would expect HER2 conversion to preferentially occur from negative to positive, conversion both ways has been described before [17, 18, 20‐23]. One mechanism of conversion from positive to negative may well be explained by clonal selection of HER2 negative cells during the metastatic process, for example, elicited by trastuzumab therapy [39].

When considering the different metastatic sites, receptor conversion seemed to occur mostly in liver and brain metastases, but only for PR conversion this was significant. The reason for this observation is unclear, and these results need to be interpreted with caution.

One limitation to receptor conversion studies is the lack of internal control cells in samples from most of the metastatic sites. An exception to this is the liver that bears ERα and to a lesser extent PR expression in hepatocytes. Since most receptor conversion was seen in the liver, and the fact that most biopsies from breast cancer metastases are small and therefore probably quickly and well fixed, it is unlikely that these issues play an important role. A further limitation was the deliberate choice not to include the preferential metastatic site of breast cancer: the bone (marrow). This was to avoid false negative results due to decalcification artefacts. Such false negative results are not easy to trace since internal positive control cells in the bone marrow are lacking. Future studies selectively studying small bone biopsies that were not decalcified may shed further light on percentages of conversion on this metastatic site.

Nearly 11% of the patients converted from ER+ or PR+ to ER-/PR- and 3.4% from ER-/PR- to ER+ or PR+ (using the 10% threshold); in these cases steroid receptor conversion could be especially clinically relevant.

Together with the HER2 conversion rate of 5.2% by IHC, in about 19% of metastatic patients the choice of systemic therapy is suboptimal when solely based on IHC of the primary tumor. However, before concluding that metastases should be biopsied when possible, there are a few issues to consider. Clinicians would probably be inclined to treat patients with positively converted distant metastases (3.4 + 2.6 = 6%) with the matching (hormonal or trastuzumab) systemic treatment, but for patients with negatively converted distant metastases this is probably more complicated. First, technical problems in cases with receptor negative metastases cannot be fully excluded. Second, there may be heterogeneity between distant metastases from which only one may get biopsied [40]. Third, there are few clinical data on response to systemic treatment in negatively converted patients. Therefore, clinicians might be inclined to consider (hormonal or trastuzumab) systemic treatment even in negatively converted patients. Nevertheless, when a biopsy of a distant metastasis is available, hormone receptor and HER2 status should be reassessed in these biopsies and the tests results should be critically evaluated in conjunction with ER, PR and HER2 status of the primary tumor. In the future, non-invasive assessment of the receptor status by molecular imaging may form an alternate and more functional way of assessing receptor status of distant metastases [24‐26], especially for metastases at inaccessible sites, also providing information on heterogeneity of receptor status between distant metastases.

In conclusion, receptor conversion in distant non-bone breast cancer metastases indeed occurs, is relatively uncommon for ERα and HER2, more frequent for PR, and seems to be more frequent in liver and brain metastases. In a considerable number of patients such conversion could theoretically have consequences for the systemic therapeutic regimen. For this reason, receptor status should therefore be reassessed on available biopsies from distant metastases. Whether distant breast cancer metastases should be more routinely biopsied when possible, will likely be the subject of further discussion. In the future, non-invasive assessment of the receptor status by molecular imaging may be an attractive alternative, especially for metastases that are difficult to biopsy.