In this study we describe a robust model of surgery-accelerated metastasis in OS, allowing for the mechanistic analysis of this effect. We demonstrate that surgical excision of the primary tumor-bearing limb resulted in significant increases in the number of gross metastatic nodules, the number of micrometastatic foci, overall metastatic burden, and effect-size consistent increases in the average size of each metastatic focus. Surprisingly, subjecting animals to equivalent surgical stress through amputation of the contralateral non-tumor bearing-limb, leaving the primary tumor in place, did not produce the same enhancement of metastasis. This strongly indicates that surgery-accelerated metastasis is mediated, at least in part, by the absence of the tumor and not just factors intrinsic to surgical wounding. Additionally, we implicate macrophages as key mediators in this effect by demonstrating that surgical excision of the primary tumor produces persistent changes in macrophage phenotype within the metastatic niche to that of a more pro-tumor state. Finally, we provide evidence that utilizing gefitinib as a peri-operative adjunct to surgery can effectively mitigate the harmful effects of surgical excision of the primary tumor by preventing tumor supportive phenotypic changes in macrophages, decreasing metastatic burden, and enhancing survival.
Although surgery-accelerated metastasis has been repeatedly described, the underlying mechanisms that contribute to this effect remain poorly understood, and as a result there is a paucity of targeted treatments. Advancements in the study of surgery-accelerated metastasis have been limited in significant part due to the difficulty of developing experimental animal systems that can accurately recapitulate the human condition. In the model of surgery-accelerated metastasis in OS used in this study, we directly seed the lungs with micrometastatic foci through an intraosseous injection producing both a primary tumor that can then be manipulated, as well as pulmonary micrometastases that can be subsequently analyzed . Clinically, we know that the majority of OS patients have micrometastatic pulmonary disease at the time of primary tumor resection, thus by conducting surgery 1 week after micrometastases are present in the lung we are able to study the effects of surgery on the later steps of metastatic progression, recapitulating this clinical scenario.
We also examined the effects of surgery on survival in this K7M2-BALB/c model of OS. We found that even though surgical excision of the primary tumor may produce a small reduction in median survival compared to tumor-bearing controls, this surgical intervention did result in a small number of long-term survivors compared to the tumor bearing group (2 animals vs 0 animals, respectively). These data suggest that despite the pro-metastatic effects of surgery, there may be benefits to removal of the primary tumor, possibly caused by morbidity of the growing primary tumor itself. Consistent with this are the findings of the study by Rashid et al. , which demonstrated that although removal of the primary tumor in a murine model of metastatic breast cancer resulted in accelerated growth of metastatic lesions, overall survival was improved by primary tumor resection. Furthermore, the authors demonstrate that when primary tumor resection decreased overall tumor burden substantially, accelerated growth of metastatic lesions did not increase overall tumor burden compared to the no-surgery group and survival was subsequently improved, which was not the case when primary tumor resection did not significantly reduce overall tumor burden . Our observations are also consistent with the findings of Simpson-Herren et al.  in a mouse model of Lewis lung cancer, where surgical excision of the primary tumor resulted in a significantly increased growth rate of lung metastasis, resulting in a small, but consistent, decrease in median survival. Furthermore, in a study by Dillekås et al. examining the recurrence patterns of women with breast cancer after mastectomy over a 30 year period, it was found that while breast surgery was clearly an independent stimulating event for the growth of metastases and increased relapse, it did not result in worse long-term disease-free survival . Taken together these data indicate that despite some of the harmful effects of surgery, the benefit of surgical procedures in diagnosis, treatment, and cure of oncologic disease is generally considered to be without argument . The importance of demonstrating that there are in fact unintended tumor-promoting effects of surgery is that we can now focus on understanding the mechanisms of these effects in order to produce targeted therapies to further augment the benefits of surgery.
While there are many hypotheses as to the underlying mechanisms contributing to surgery-accelerated metastasis, one of the most pervasive is that surgical wounding generates a permissive tumor environment through alterations in immune function [4, 6,7,8,9,10,11, 18, 32]. Increased macrophage infiltration of primary tumors outside the context of surgery has been shown to portend a poor prognosis [33, 34]. Functionally, TAMs appear to be predominantly alternatively activated macrophages exhibiting immunosuppressive and pro-tumorigenic functions. However, conceptualizing macrophages strictly as either pro-tumor or anti-tumor types is an over simplification of their complex biology and plastic nature. While generally TAMs appear to be tumor supportive by secreting pro-angiogenic, pro-growth, and immunosuppressive factors, there are in actuality a multitude of different macrophage types associated with the tumor.
Few studies have looked at changes in TAM populations after surgery. Krall et al.  demonstrated in a mouse model of breast cancer that surgical wounding results in a systemic upregulation of macrophages, increasing their availability to be recruited into tumors. Furthermore, that study demonstrated that the macrophages that were recruited into tumors following surgical wounding showed increased expression of CD206 on their surface, a marker of pro-tumor macrophage function . Similarly, Tham et al. found that following resection of primary melanoma in a mouse model there is increased infiltration of macrophages both within primary tumor recurrences and within metastases, and that this macrophage infiltration is associated with increased proliferation of tumor cells. However, the phenotype of the macrophage infiltrate within the metastases was not examined in their study . Interestingly, Maloney et al.  also reported that the proliferative capacity of K7M2 OS cells was enhanced when co-cultured with bone marrow-derived macrophages in an in vitro assay. Although the exact phenotype of the macrophages used in that assay was not characterized, those macrophages were grown in media containing M-CSF, a known driver of a pro-tumor macrophage phenotype . In this current study we have shown a trend toward an increase in the average size of metastatic foci following surgical resection, which was associated with a shift in macrophages to a pro-tumor phenotype, suggesting that pro-tumor macrophages can enhance proliferation in OS as well.
To our knowledge this is the first report demonstrating that surgical resection of the primary tumor induces a sustained shift in macrophage phenotype within the metastatic niche to a pro-tumor state, correlating with enhanced metastatic outgrowth in OS. The focus of our study was specifically on the alterations to macrophage function within the metastatic niche following surgical resection, and how this contributes to surgery-accelerated metastasis. However, we cannot exclude the role of other post-surgical physiologic changes in metastatic enhancement. Some of these proposed mechanisms include release of immunologic and neuroendocrine factors such as prostaglandins, catecholamines, and glucocorticoids, creation of a systemic pro-angiogenic environment, and release of wound-healing related growth factors [5, 6, 32]. We also cannot exclude the effect of changes to other immune cell types within the metastatic niche such as T-cells or NK cells, or the effect of changes to myeloid cells in circulation after surgery. Examining how these other immune cell types change following surgery and potentially contribute to surgery-accelerated metastasis is a natural continuation of our work and will be the focus of future studies. It is most likely that multiple mechanisms act in conjunction with one another to ultimately support tumor recurrence and metastatic outgrowth after surgical intervention. However, pro-tumor macrophages may contribute to the other proposed mechanisms as well, as pro-tumor macrophages themselves secrete angiogenic factors, growth factors, matrix-remodeling proteins, cytokines, and chemokines [4, 14, 32]. Thus, by investigating post-surgical changes in macrophages we may garner a better understanding as to how the mechanisms underlying surgery-accelerated metastasis may be linked, to which this report contributes. Currently, we are engaged in further studies to correlate the post-surgical changes in macrophage phenotypic markers identified in this study with functional changes in these macrophages by using PCR analysis. We plan to examine macrophage gene expression changes following surgery in genes specifically related to cancer and immunity crosstalk, including genes related to angiogenesis, matrix-remolding, and growth factor production. This will be the work of a future study.
The design of the surgical model used in this study was such that the entire tumor bearing limb of the mouse was amputated, as limb sparing surgery in a mouse model is technically prohibitive. This model was chosen to mimic as closely as possible the clinical situation of total resection of the primary tumor with pre-existing pulmonary micrometastases. Despite its benefits, the orthotopic model of osteosarcoma used in this study has some important limitations that should be noted. As discussed previously by other authors, injection of a tumor cell suspension into the murine tibia is a technically difficult process due to its curved anatomy and the limited volume of the tibial marrow cavity [35, 36]. These technical limitations of the model, along with intrinsic variations in cells and animals, can lead to inter-experimental variation in regards to the size of primary tumor generated and the number of metastases created within the lung. However, although the absolute number of metastases present within the lungs differed between experiments, the rise in metastases following surgical excision of the primary tumor occurred consistently, as did the inhibition of surgery’s effect on metastasis with peri-operative treatment with gefitinib. Additionally, our study is limited by the fact that it was conducted with only one OS cell line and within one mouse strain, and it is possible that the contribution of macrophages to surgery-accelerated metastasis may differ between OS cell lines and in mice of different genetic backgrounds. This will be focus of future studies. We note however, that macrophages have been implicated in surgery-accelerated metastasis in several different tumor types, including breast cancer and melanoma [11, 18].
In order to examine the distinct effects that surgical wounding and removal of the primary tumor may have on surgery-accelerated metastasis, we included a group in our experiments where the contralateral non-tumor-bearing limb was amputated. This subjected animals to the equivalent surgical stress of an amputation, while leaving the primary tumor in place. Interestingly, amputation of the non-tumor bearing limb did not result in the same increase in metastatic nodules as amputation of the primary tumor bearing-limb. In the acute post-operative period, surgical intervention regardless of removal of the primary tumor resulted in a shift toward a pro-tumor macrophage phenotype within the metastatic niche. We hypothesize that the magnitude of the surgical insult incurred by amputation is so significant that it may mask small differences in macrophage phenotype between the two groups. Long-term however, macrophage phenotype in the contralateral amputation group returned to near that of non-operated controls, while those animals that underwent surgical removal of their primary tumor showed a persistently altered pro-tumor pulmonary macrophage phenotype. This suggests that enhanced metastatic outgrowth and persistent tumor-supportive macrophage phenotype within the metastatic niche is specifically related to removal of the primary tumor, and is not solely dependent on factors intrinsic to surgical stress.
As reviewed by Chiarella et al. , the ability of the primary tumor to exert a controlling and inhibitory effect on the growth of distant metastases is known as concomitant tumor resistance, and may be a contributory mechanism in surgery-accelerated metastasis. It has been postulated that the primary tumor can inhibit distant metastases either directly through the production of anti-angiogenic and anti-proliferative substances or indirectly by generating an anti-tumor immune response that is capable of inhibiting small micrometastases . Concomitant tumor resistance has been demonstrated in a number of tumor types including sarcomas. In mouse models of both melanoma and sarcoma, Schatten et al.  demonstrated that surgical resection of the primary tumor led to an increase in both the frequency and size of pulmonary metastases compared to the no-surgery group or the group that underwent amputation of the non-tumor bearing limb. Similarly, in a model of osteosarcoma, Tsunemi et al.  demonstrated that resection of the primary tumor produced an increase in pulmonary metastasis compared to a sham-surgery group, and the increase in metastasis was associated with a systemic reduction in the angiogenesis inhibitor endostatin, resulting in systemic enhancement of angiogenesis. The mechanism by which the primary tumor in our model of OS modulates macrophages to inhibit metastatic growth while in place is unclear, and will be the focus of future work.
Our work and the work of others suggests that removal of the primary tumor may contribute to surgery-accelerated metastasis. However, other studies demonstrate that surgical wounding itself is sufficient to provoke metastatic out growth. Krall et al. successfully create an experimental model of breast cancer where primary tumor resection and surgical wounding are uncoupled. In that report, surgical wounding itself was able to generate a systemic response that led to the outgrowth of tumors at distant sites . Similarly, Hobson et al.  demonstrated the acute inflammation generated by biopsy alone was sufficient to increase the frequency of pulmonary metastases in a mouse model of breast cancer. We demonstrate that removal of the primary tumor is needed for surgery-accelerated metastasis in OS; however, these studies underscore the complexity of the mechanisms contributing to surgery-accelerated metastasis. Clinically, tumor resection and surgical wounding occur jointly, however separating these two processes experimentally aids in providing important mechanistic insights.
Although there is a body of work, including this report, that surgical removal of the primary tumor is capable of generating systemic effects that result in accelerated metastasis, we certainly do not suggest that surgical resection be abandoned in any way. To the contrary, our findings demonstrate the importance of continued study in this area to further understand the mechanisms behind surgery-accelerated metastasis. Understanding these mechanisms will allow for the development of novel treatment approaches that will serve to further enhance the benefits of surgery. The findings presented here provide rationale for future work on the mechanisms that govern the pro-tumor macrophage phenotypic alterations that occur after surgical resection and how those alterations contribute to enhanced metastasis. Importantly, this study identifies gefitinib, an already FDA-approved medication, as a potential surgical adjunct to mitigate the unwanted consequences of surgery, thereby enhancing its benefits.