Abstract
There is often a pressing need for reconstruction after cancer surgery. Regenerative therapy holds the promise of more natural and esthetic functional tissue. In the case of breast reconstruction postmastectomy, volume retention problems associated with autologous fat transfer could be ameliorated by augmentation with cells capable mediating rapid vascularization of the graft. Intentional placement of regenerating tissue at the site of tumor resection raises questions concerning the possibility of promoting cancer recurrence. Here we review coculture and animal models of tumor/mesenchymal stem cell interactions under regenerating conditions. Available evidence from case reports, cell lines, and clinical isolates favors the interpretation that regenerating tissue promotes the growth of active, high-grade tumor. In contrast, dormant cancer cells do not appear to be activated by the complex signals accompanying wound healing and tissue regeneration, suggesting that engineered tissue reconstruction should be deferred until cancer remission has been firmly established.
Introduction
Cancer surgery can be disfiguring
Autologous fat transfer
Autologous fat transfer (AFT) for soft tissue reconstruction was initially described more than a century ago in a German-language article entitled Fettransplantation (Fat Transplantation). The initial indication for fat transplantation was for correction of facial defects 1 and was soon after introduced for breast reconstruction postmastectomy. 2 However, fat injection into the breast became controversial among plastic surgeons because of potential complications such as local calcifications and interference with mammographic breast cancer surveillance. 3 AFT remains an attractive reconstructive technique with low complication rates. 4 The first successful application of soft tissue regenerative therapy after cancer was performed more than a century ago by Czerny, 2 who restored breast symmetry postmastectomy by transplanting a benign autologous lipoma. Although transplantation of autologous fat yields satisfactory short-term cosmetic results, volume retention has been a recurring problem.5,6 Recent fat transfer and lipoinjection protocols have focused on the addition of autologous adipose-derived stem/stromal cells (ASC) or freshly isolated adipose stromal vascular cells to promote graft volume retention.7–9 Enthusiasm for the use of stem cell-augmented adipose transplantation for breast reconstruction has been tempered by the fear that the transplantation of self-vascularizing, self-renewing adipose tissue may promote tumor recurrence by supporting reactivation of occult breast cancer cells.
Potential advantages of cellular therapy
The variability in long-term graft survival6,10 has been attributed to differences in local angiogenesis.11,12 Recently, the addition of ASC to whole fat grafts was proposed to support the formation of a new vasculature8,13 and promote graft retention. 14 Fat tissue is a rich source of both endothelial progenitors15,16 which can mediate angiogenesis and multipotent MSC. 17 Both populations are present in the freshly isolated adipose stromal vascular fraction (SVF). 18 The SVF, when expanded in short term culture,17,19,20 has been termed ASC 21 and resembles MSC in many important respects, which will be discussed below. The rationale for combining whole fat or lipoaspirate with SVF cells or ASC is that the organized fat tissue may serve as a scaffold upon which more concentrated stem/progenitor cells can organize and differentiate. Another attractive approach for guiding the three-dimensional organization of engineered tissue reconstruction is through the use of scaffolding,22–24 which may be mineral, synthetic polymers, or biological and may incorporate growth factors naturally 25 or by design.26–28
Models of tumor cell/MSC interactions
Therapy for epithelial cancers is rarely curative and late recurrence after apparently successful therapy provides prima facia evidence for the persistence of dormant cancer cells. Precisely how regenerating engineered tissues may interact with active and dormant cancer cells in vivo is currently unknown. However, interactions of proliferating ASC and MSC with tumor cells have been addressed in coculture and human/murine xenotransplantation models. The major contribution of these studies is that they have provided insight into mechanisms of tumor invasion, demonstrating that secretion of the chemokine CCL5 by bone marrow-derived MSC (BM-MSC) 29 or ASC increase the motility of breast cancer cell lines in in vitro models of tumor invasion. These findings were reproduced by the addition of exogenous CCL5 to breast cancer cell line cultures. 30 Further, BM-MSC have been shown to promote in vitro epithelial to mesenchymal transition of breast cancer cells and reduce expression of proliferation-associated genes. 31
Because these studies utilized immortalized cancer cell lines that grow rapidly in culture and give rise to large tumors in a very short time, they fail to model the crucial aspects of tumor heterogeneity, tumor dormancy, and reactivation of occult tumor cells. For example, coculture of MDA-MB-231, a hormone-independent breast cancer cell line, with MSC resulted in greater cell expansion after 4 days in culture (1.7-fold more cells), but tumor cells alone had a doubling time of ∼24 h. 32 This extreme lack of clinical realism is also apparent in tumor transplantation models. For example, Yu et al. coinjected 1 × 106 each of H460 cells (a human lung cancer cell line) and human ASC and reported measurable tumors in 5 days that grew to 20 mm3 in 10 days. 33 Muehlberg et al. injected 5000 4T1 cells (a murine breast cancer cell line) and observed 125 mm3 tumors in 21 days. Addition of murine ASC increased the size of 21 day tumors to 400 cubic mm. 34 In the model reported by Karnoub et al., tumors ranging in size from 50 to 500 mm3 were detectable by day 30. 29 Clearly if these models bear any relevance to human disease, it would be in the context of rapidly growing high-grade therapy unresponsive tumors, where reconstructive surgery would not be a consideration. Our own approach has relied on a model system utilizing unpassaged sort-purified clinical isolates injected in limiting numbers (100 cells).35,36 This model requires 3–6 months for the generation of vascularized epithelial tumors in the size range of 5–10 mm3.
Apart from the ability to promote the growth of existing tumors, there is also a report in which implantation of ceramic scaffold charged with syngeneic murine MSC gave rise to sarcomas in a proportion of treated mice. 37
There is also a degree of contradiction in the literature, in which ASC or MSC have been shown to inhibit the growth of tumor cells. For example, ASC have been reported to inhibit tumor growth of MDA-MB-231 cells, 38 and induce cell death of pancreatic adenocarcinoma cells lines, hepatocarcinoma, colon cancer, and prostate cancer. 39 MSC reportedly inhibit tumorigenesis from Kaposi's sarcoma cells 40 and hepatoma cells, 41 bone metastasis of prostate cancer, 42 and in vitro growth of hepatoma, lymphoma, and insulinoma cell lines. 43 Similarly, MSC derived from human umbilical cord blood are cytotoxic for human malignant glioma cells. 44 Multiple mechanisms have been invoked to relate these findings to the cytokines, chemokines, and prostaglandins secreted by MSC, 39 but little experimental evidence has been provided to define the mechanisms of tumor inhibition. Khakoo et al. 40 proposed contact-dependant inhibition of Akt protein kinase, whereas Qiao et al. cited down regulation of Wnt signaling, c-Myc, and Bcl-2. 41
Phenotypic and functional characteristics of ASC and MSC
ASC phenotypically resemble BM-MSC 19 and share their multipotentiality and many surface markers17,45 (Table 1), but are orders of magnitude more prevalent in fat than BM-MSC are in BM aspirates. In disaggregated freshly isolated adipose tissue, phenotypically defined pericytes comprise 1.7%, whereas CD34+ supra-adventitial stromal cells represent 27% of nucleated cells. 18 In contrast MSC represent only 0.001%–0.004% of BM aspirate cells.20,46 Although ASC and BM-MSC are both promising candidates for reconstructive cellular therapy after tumor resection, the potential risk of promoting tumor reactivation is controversial. This is especially germane in breast cancer considering that up to 20% of patients will suffer from cancer recurrence during the first decade after adjuvant therapy. 47
Multipotentiality demonstrated at the clonal level.
ASC, adipose-derived stem cell; BM, bone marrow; MSC, mesenchymal stem cell.
These data indicate strong phenotypic and functional similarities between ASC and BM-MSC. Like BM-MSC, ASC are able to give rise to differentiated progeny with characteristics of bone, cartilage, fat, and vessels and, like MSC, were predominantly CD105+/CD73+/CD90+/CD44+ (Table 1). However, low passage ASC also contain minor populations expressing the adipose pericyte-associated marker CD146 18 and CD34, a marker not associated with BM-MSC. The in vitro secretomes of ASC and MSC are also similar (Table 2). However, ASC secrete significantly higher quantities of leptin (140-fold) and adipsin (20-fold), and lower quantities of the angiogenic factors vascular endothelial growth factor and soluble vascular cell adhesion molecule (sVCAM). Production of leptin and adipsin is characteristic of mature adipocytes, but has also been described in ASC cultures. 35 The hormone leptin has been reported to increase angiogenic and proliferative signaling in ER+ and ER− breast cancer cell lines. 48 Adipsin, a trypsin-like serine protease increases the concentration of acylation-stimulating protein, another adipose-derived hormone that regulates triglyceride synthesis. 49 It has no known role in cancer.
BM-MSC data are unpublished internal control data of the authors using a single MSC isolate at passage 3. Analytes were measured by Luminex assay of supernatants harvested 3 days after plating when cultures were ∼80% confluent. All supernatants (MSC and ASC) were processed simultaneously. The methods, media, standards, and blanks were identical to those published for ASC 35 . Bolded rows show major differences.
CRP, C-reactive protein; IL, interleukin; PAI, plasminogen activator inhibitor; TGF, transforming growth factor; TNF, tumor necrosis factor; VEGF, vascular endothelial growth factor.
In vitro model of ASC interaction using clinical isolates
In vitro model systems can detect effects of secreted growth factors and cell adhesion-mediated effects, but lack the complexity of xenograft models. Through an in vitro coculture model utilizing heterogeneous cells isolated from clinical isolates (malignant pleural effusions), we confirmed the enhancing effect of ASC on the proliferation of breast cancer cells 35 which appeared as nest of epithelioid cells among a monolayer of carboxyfluorescein succinimidyl ester (CFSE)-labeled stromal cells. Our results are in agreement with those studies showing that the presence of mesenchymal cells promotes the growth of highly proliferative tumor cells.32,50
Tumor cell heterogeneity
The epithelial component of breast cancer clinical isolates is heterogeneous. Initially, a tumor cell subset identified as CD44+/CD24-/CD326 (ESA, EpCAM)+ was shown to be enriched for tumorigenic cells as detected in a xenotransplantation model. 51 In an accompanying editorial it was hypothesized that these cells represent breast cancer stem cells. 52 Gradually, evidence has accrued substantiating the fact that clonogenic tumor cells share, constitutively or conditionally, many characteristics with adult tissue stem cells. Most prominently these include self-renewal53,54 and therapy resistance.36,55 As a cause or consequence of these functional similarities, adult tissue stem cells and clonogenic epithelial tumor subsets share expression of several markers. We previously demonstrated the tumorigenicity of CD90+ breast cancer cells separated on the basis of light scatter into resting (low light scatter) and active (high light scatter) populations. Small resting CD90+ cells gave rise to tumors with high efficiency (50–100 cells/injection),36,56 whereas large, active CD90+ cells were tumorigenic at high (600–13,000 cells), 36 but not low 56 dose. We further explored the phenotypic differences between resting and active CD90+ tumor cells, where low light scatter resting CD90+ nonheme cells were mostly quiescent in contrast to their high light scatter counterpart, which had a higher proportion of cycling/aneuploid cells. 56 We have demonstrated that ASC failed to augment the tumorigenicity of small resting CD90+ tumor cells, whereas they markedly enhanced tumorigenesis mediated by active CD90+ tumor cells. 35 Dormant and proliferating breast cancer cells display distinct genome-wide expression signatures, including differences for a high number of angiogenesis-related genes. 57 This is consistent with the hypothesis that ASC may support survival and proliferation of tumor cells in vivo by promoting angiogenesis, to which active cells are preferentially receptive.
Discussion
Escape from tumor dormancy: A working hypothesis
Throughout this article we have drawn a distinction between two potentially tumorigenic populations, resting (dormant) and active (proliferating) tumor cells. This distinction has parallels with adult tissue stem cells versus transit amplifying populations. The notion of cancer dormancy is prevalent in the cancer literature, but ill defined. Dormant cancer is subclinical cancer, and is known because of tumor recurrence after a symptom-free interval. It is not known whether dormant tumor cells are out of cell cycle (i.e., G0), or persisting in a dynamic state of balanced proliferation and death. The same can be said for normal tissue stem cells. 58 Regardless of the mechanism by which subclinical tumor persists, it is useful to hypothesize that dormancy is an intrinsic characteristic of the resting tumor cell, perhaps imposed by epigenetic programming. 59 As such, transition between dormant and active states requires genetic reprogramming and not merely the presence of signals such as those provided by hormones and hormone receptors, or even the presence of mutations that bypass the need for such signaling. Whatever the stimuli that drive dormant stem-like cancer cells into an active tumorigenic state, we hypothesize that they are distinct from those that favor the survival and proliferation of active progenitor-like tumor cells (Fig. 1). Therefore, the introduction of pro-angiogenic autologous mesenchymal stem cells to the site of a tumor bed would be non-contributory to local recurrence. Indeed, if increased local vascularization at the site of a tumor bed was an independent risk factor for recurrence, this fact would have become clear after observing patients reconstructed with local tissue flaps. In this commonly accepted method of reconstruction, a well vascularized tissue flap is in direct contact with the tumor bed and extensive local vascular remodeling occurs. Immediate autologous tissue flap reconstruction has not been correlated with increased recurrence rates. 113

Working hypothesis of the interaction between regenerating tissue and an epithelial cancer. The effect of adipose-derived stem cell administration on active breast cancer is depicted here. Introduction of ASC, which self-replicate and give rise to adipose, vessels, and stroma, promotes the growth of active tumor (purple) and induces motility and upregulation of adhesion molecules, promoting invasion and metastasis (deep purple). Dormant cancer cells, shown here nested in a normal breast duct (hot pink, inset), are unaffected by the wound healing signals resulting from regenerative therapy. ASC, adipose-derived stem cells. Color images available online at www.liebertonline.com/ten.
Risks assessment
In the absence of substantial clinical experience with tissue-engineered reconstructive surgery after cancer, one can also look to biological parallels to estimate the risks imposed by intentionally placing regenerating tissue at the site of tumor resection. The first and most obvious parallel is the wound healing that normally accompanies cancer therapy of any kind, whether surgical, chemotherapeutic, or radio-ablative. Even before antineoplastic therapy has been initiated, the tumor microenvironment has much in common with a wound. 60 The similarities have been well described 61 and include the presence of inflammation, growth factors, cross-linked fibrin, fibroblast activation, angiogenesis, and the deposition of a network of extracellular matrix. Thus, it is difficult in most cases to determine the extent to which wound healing associated with treatment contributes to relapse at the primary tumor site. Cancer recurrence in mastectomy scars has been reported, but is rare. 62 In colon cancer, recurrence at scar sites is also infrequent and associated with aggressive systemic disease. 63 There are even case reports of high-grade tumors recurring along the tracks of laparoscopy.64–66 In stage I/II invasive breast cancer patients treated with radiation, in whom the incidence of early local recurrence is very low, high mitotic activity and high tumor grade were the major predictors of local recurrence. 67 Thus, the common feature shared by cancers that are recruited into treatment-associated wounds appears to be the presence of aggressive active disease.
A second, less obvious parallel comes from experience with allogeneic and autologous hematopoietic stem cell transplantation for hematologic malignancies, where there are decades of experience in regenerative therapy for cancer.68,69 When disease is active at the time of transplantation (i.e., the patient has had multiple remissions or has active disease at the time of myeloablative therapy), relapse often occurs early in the peritransplant period when BM regeneration (and associated reestablishment of the components of the BM microenvironment) is at its peak. 70 When this occurs, the recurrent disease is usually very aggressive. In contrast, patients transplanted in first remission often have extended disease-free survival. The proportion of patients who eventually relapse often do so years after transplant, indicating that (1) residual disease survived myeloablative therapy; (2) these cells were dormant; and (3) dormant residual malignant cells were not reactivated as a result of massive injury to the BM and explosive hematopoietic regeneration ediated by the graft. An analogous argument can be made for cytokine mobilization in the context of autologous hematopoietic stem cell transplantation, where patients in remission are treated with consolidation chemotherapy and the hematopoietic growth factor granulocyte colony stimulating factor (G-CSF). 71 The patients experience BM suppression followed by explosive regeneration without triggering early relapse.
In the 1990s autologous hematopoietic transplantation was widely used to rescue the BM of breast cancer patients undergoing dose-intensive chemotherapy. This practice was discontinued when randomized trials failed to show superiority over conventional therapy. 72 However, the massive damage to all proliferative tissues by high-dose cytotoxic therapy and the ensuing tissue regeneration (including marrow reconstitution, revascularization, and reepithelialization) did not promote breast cancer relapse.
Currently, there is relatively little clinical experience pertaining specifically to regenerative therapy in the context of epithelial cancers. Delay et al. published a series of 880 breast reconstructions with autologous fat alone performed by injection of lipoaspirate. Some of the participants included patients undergoing nipple-areola reconstruction after mastectomy. In this cohort, which had a maximal follow-up of 10 years, there was no detectable increase in the risk of local recurrence or new cancer development. 4 A similar series was reported by Illouz and Sterodimas, who performed AFT on 820 patients, 381 of whom were treated for asymmetry after mastectomy and breast reconstruction. 73 Although 230 patients were followed with yearly mammography and ultrasonography for an average of 11.3 years, the author made no mention of the incidence of locoregional recurrence or metastasis among the breast cancer patients. Yoshimura et al. reported on the use of SVF-augmented AFT for cosmetic breast augmentation in 40 women, none of whom experienced serious complications during a follow-up interval ranging from 6 to 42 months. 8 Among the adverse effects reported after SVF-augmented breast reconstruction were calcifications and cyst formation in 4 of 40 patients. An additional cohort of 15 patients had successful breast reconstruction after experiencing complications of breast implant surgery. 74 In a cohort study (mean 7.2 year follow-up) reported by Rigotti, 137 patients treated with AFT breast reconstruction did not show increased risk of local recurrence after this treatment. While this study lacked a formal control group (the recurrence rates after treatment were referenced to both historical data and pre-reconstruction recurrence rates for the cohort), the data suggest that the addition of autologous adipose cells to a tumor bed does not impact any nascent cancer cells. 114 Yoshimura et al. also performed cell-augmented AFT on eight patients undergoing breast reconstruction after mastectomy, 8 but the clinical status of these patients was not reported.
Taken together with our published results in a xenograft model, 35 which indicate that ASC augment the growth of active but not resting breast cancer cells, the available data suggest that the critical factor determining whether regeneration augments tumor growth is the state of residual tumor: active disease is promoted, whereas dormant tumor is insensitive. This suggests that reconstructive therapy utilizing ASC-augmented whole fat should be deferred until cancer remission has been firmly established.
Footnotes
Acknowledgments
This work was supported by grants BC032981 and BC044784 from the Department of Defense, grants R01CA114246 (JPR) and NO1 HB 37165 from the NIH, the Hillman Foundation, the Glimmer of Hope Foundation. and the Commonwealth of Pennsylvania, through the McGowan Institute of Regenerative Medicine. Vera S. Donnenberg is a CDMRP Era of Hope Scholar.
Authors Contributions
Vera S. Donnenberg and Albert D. Donnenberg and Ludovic Zimmerlin, graduate student in the Donnenberg laboratory, contributed equally to this article. J. Peter Rubin provided clinical perspective and expertise in AFT.
Disclosure Statement
None of the authors have competing interests that would have influenced the preparation of this article.
