Aerosol Gemcitabine after Amputation Inhibits Osteosarcoma Lung Metastases but Not Wound Healing

Background In newly diagnosed osteosarcoma (OS) patients, the time between surgery and resumption of chemotherapy is 2–7 weeks. Delays > 16 days are associated with increased risk of relapse and decreased overall survival. Identifying an effective therapy that can be used postoperatively may prevent relapse. We investigated whether aerosol gemcitabine (GCB) initiated after tumor resection inhibited the growth of OS lung metastases without affecting the wound-healing process. Methods Mice were injected intratibially with OS cells. Amputation was performed when the tumor reached 1.5 cm. Full-thickness excisional wounds were also made on the dorsal skin and tail. Aerosol GCB or PBS was initiated 48 hours after amputation (3 times/week for 3 weeks). Wound sections were evaluated by immunohistochemistry for Ki-67 (proliferation), CD31 (vessels), VEGF, IL-10, bFGF, mast cells, macrophages, and M1/M2 macrophage ratios. The lungs were analyzed for macro- and micrometastases. Results Aerosol GCB inhibited the growth of the lung metastases but had no effect on the 3 phases of wound healing in the dorsal skin, tail, or bone. Production of cytokines at the wound sites was the same. Conclusion These data indicate that initiating aerosol GCB postoperatively may kill residual lung metastases thereby preventing relapse and improve survival.


Introduction
Osteosarcoma is the most common malignant bone tumor in both adults and children. While the introduction of combination chemotherapy given in addition to surgery improved the overall survival from 20 to 65%, survival rates have remained stagnant for >25 years [1][2][3]. e lung is most common site of metastatic spread, and the majority of newly diagnosed patients have undetectable microscopic lung metastases at the time of diagnosis providing the rational for giving preoperative and postoperative chemotherapy. Despite the use of aggressive combination chemotherapy, the development of visible lung metastases continues to be the major challenge in curing this disease. For patients who develop relapsed disease in the lung, the 5-year survival rate is only 20% [4,5], and second-line chemotherapy has not made a signi cant impact on improving this outcome [6]. It is therefore critical to eradicate these micrometastases during the initial treatment period.
Prior to surgery, systemic chemotherapy serves to control the growth and size of the primary tumor to decrease the morbidity of surgery in addition to targeting the microscopic disease in the lung. Chemotherapy is stopped a few weeks before and for several weeks after surgery due to the interference of systemic chemotherapy on wound healing. Once patients recover from surgery, chemotherapy is resumed to prevent macroscopic metastases from developing.
is results in a break or delay in chemotherapy administration, leaving patients untreated for a signi cant period of time when tumor rebound is possible. It is therefore important that this delay be as short as possible. e standard time between the surgery and the resumption of chemotherapy treatment is 2-4 weeks, but delays as much as 6-8 weeks have been reported in patients with postoperative complications such as infection or other comorbidities that lead to delayed healing [7][8][9]. Delays > 16 days have been associated with an increased risk of relapse and a signi cant decrease in overall survival [7,8]. With 21-40% of patients experiencing delays [7,8], identifying an e ective "bridge therapy" against osteosarcoma lung metastases which does not interfere with postoperative healing and can be used in the immediate postoperative period has the potential to prevent relapse and increase both the event-free and overall survival.
Wound healing is complex and includes 3 distinct phases [10,11]. First is the in ammatory phase which consists of the recruitment of in ammatory cells, including neutrophils, mast cells, and macrophages, both M1 and M2. M2 macrophages promote cell proliferation and tissue repair, while M1 macrophages inhibit cell proliferation. e second phase involves the migration and proliferation of broblasts and endothelial cells [12,13]. Tissue remodeling is the third phase. Several speci c cytokines and growth factors are critical to the healing process including IL-10 and bFGF [14][15][16][17]. A potential bridge therapy candidate should not interfere with any of the three wound-healing phases.
Gemcitabine (GCB) is a deoxycytidine analogue that causes DNA damage, initiating cell death [18]. GCB given systemically together with Taxol has shown modest activity against relapsed large lung metastases [19]. e activity of aerosol delivery to target OS lung metastases has been investigated. Aerosol therapy has the advantage of delivering agents directly to the lung, the organ where micrometastases exist, resulting in increased drug concentration in the lung, decreased drug levels in the circulation, and decreased systemic toxicity [20]. Using 1/10th the systemic dose, the e ect of aerosol GCB against human and mouse OS lung metastases in vivo was demonstrated [21,22]. Maximum therapeutic activity was seen against microscopic disease. Peak serum levels following 1.0 mg/kg aerosol GCB were 200 ng/ml, signi cantly lower than the peak serum levels (700 ng/kg) following 1.0 mg/kg GCB given systemically. No liver, kidney, lung, or hematologic toxicity was observed in the mice following 5 weeks of aerosol GCB given 3 times/wk. In addition, aerosol GCB demonstrated signi cant therapeutic bene t in dogs with visible osteosarcoma lung metastases [23]. Finally, Phase I/II trials using aerosol GCB demonstrated that this therapeutic approach is safe with no signi cant toxicity in regard to organ function or hematologic e ects [24]. Taken together, these studies indicate that aerosol GCB may be a candidate for bridge therapy between the surgery and the restarting of systemic chemotherapy. However, there are no data in mice or patients on whether aerosol GCB a ects wound healing.
We therefore wished to determine whether aerosol delivery of GCB is e ective against OS lung metastases without interfering with wound healing in the skin and the bone following limb amputation. We evaluated the in vivo e ect of aerosol GCB initiated 48 hours after amputation of a primary OS tumor in the tibia on established OS lung metastases and the healing of bone, dorsal skin, and tail wounds. While aerosol GCB successfully eradicated the microscopic lung metastases, aerosol GCB had no signi cant e ect on the in vivo wound-healing process.

Animal
Model. BALB/c mice (Charles River, USA) were anesthetized with iso urane. K7M3 cells suspended in 10 µl of sterile phosphate-bu ered saline (PBS) at a density of 1 ×10 6 cells were injected into the tibia as previously described [22,25]. Local tumors were detected at 3-4 weeks after injection, and the metastases were visible by 5-6 weeks [25].

Amputation.
When tumor size reached 1.5 cm in diameter, the mouse was anesthetized with iso urane and the leg was amputated above the knee joint with a sterile blade followed by the application of surgical staples. To minimize the postoperative pain, buprenorphine (0.1 mg/kg) was given as s.c. injection every 8-12 hours. e amputation wound was monitored thereafter during the aerosol PBS and GCB treatment.

Preparation of Wound Tissue.
Mice were also used in the dorsal cutaneous and tail wound experiments one day after amputation. e mice were anesthetized with iso urane, and the dorsal cutaneous skin was shaved and wiped with 70% ethanol. Full-thickness excisional wounds were made by picking up a fold of skin using a sterile disposable 6 mm biopsy punch, resulting in the generation of one wound on each side of the midline [26,27]. A separate tail wound 2 Sarcoma 10 mm × 3 mm was made on the tail of each mouse down to the fascia on the same day as the dorsal wound. e two wounds were photographed every 3 days, and dimensions were measured every 3 days, particularly noting reepithelialization of the skin [28].
2.5. Aerosol GCB Treatment. Ten mice were treated with aerosol PBS (control) or aerosol GCB at 1.0 mg/kg in 10 ml saline. For comparison of serum GCB concentration, 10 mice were treated with intraperitoneal injection of 1.0 mg/kg GCB in 0.2 ml of normal saline 3 times weekly for 3 weeks as previously described [22]. Blood was collected in tubes containing 0.75 mg of tetrahydrouridine, a cytidine deaminase inhibitor, to prevent ex vivo metabolism of gemcitabine. Blood samples were processed to separate plasma and kept at −80°C until analysis. Serum was collected at 10 min, 30 min, and 24 hours after GCB treatment. Samples were sent to Pharmacy Pharmacology Research Department for analysis of the serum GCB level.

Histology and Immunohistochemistry.
Tissues from the amputation wound and dorsal cutaneous and tail wounds were collected on days 7, 14, and two months after the surgery and aerosol GCB treatment. Tissue sections were formalin xed and para n embedded. Decalci cation was performed for the amputation wound. Mast cell numbers were quanti ed using toluidine blue staining and counting the number of toluidine blue positive cells in 10 high-power eld (h.p.f) under a 20x objective lens. e total number of macrophages was quanti ed using anti-F4/80 antibody staining (ab6640, Abcam USA). M1 macrophages were identi ed and quanti ed using iNOS staining (PA5-16855, ermo Fisher Scienti c). Anti-mannose receptor antibody (ab 64,693, Abcam USA) was used to identify and quantify M2 macrophages. CD31 antibody (ab 28,364, Abcam USA) was used for endothelial cell staining; anti-VEGFR-2 (ab39256, Abcam USA) was used for VEGF staining; IL-10 and FGF-2 were measured using ab189392 (Abcam USA) and sc-1360 (Santa Cruz Biotechnology), respectively. Ki67 antibody (ab15580, Abcam USA) was used for cell proliferation and broblast activation; protein alpha antibody (ab53066, Abcam USA) was used for broblast staining. All the sections were incubated with the primary antibody overnight at 4°C in accordance with the protocols provided by companies.

Statistics.
To determine the signi cance between two groups, we used GraphPad Prism. e statistical analysis between two groups was performed using the unpaired Student's t-test. P > 0.05 was considered to have no significant di erence.

Serum GCB Levels following Aerosol versus Intraperitoneal
Administration. We rst compared the serum levels of GCB following aerosol or intraperitoneal administration. Serum GCB levels in the mice treated with GCB i.p. were 3.5-fold higher than those treated with aerosol GCB at 10 min and 30 min after administered ( Figure 1).

Aerosol GCB Had No E ect on Wound Healing but Inhibited Osteosarcoma Lung Metastases.
In order to mimic the clinical course of osteosarcoma, K7M3 cells were injected into the tibia. Micrometastases in the lung were con rmed by sacri cing 2 mice 3 weeks later. When the tumors measured 1.5 cm, tail and dorsal skin wounding and amputation of the a ected limb were performed. e mice were divided into 2 groups and treated with either aerosol PBS or aerosol GCB 24 hours after amputation. Aerosol therapy continued twice a week for three weeks. Healing of the tail, dorsal skin, and amputation site was monitored. Skin wounds were measured 4, 7, and 10 days after the initiation of aerosol therapy.
ere was no di erence in wound healing of the tail or dorsal skin between the aerosol PBS-and aerosol GCB-treated mice (Figures 2(a) and 2(b)). Similar to dorsal and tail wounds, aerosol GCB also did not a ect the healing of the amputation site as shown by H&E staining in 3 di erent regions (Figure 2(c)).
To determine whether aerosol GCB inhibited lung metastasis, ve mice from each treatment group were sacri ced after 3 weeks of aerosol therapy. Aerosol GCB signi cantly decreased the number of visible lung metastases compared to the aerosol PBS group (Figure 2(d)). Lung weights and the number of micrometastases were also signi cantly reduced in the aerosol GCB group (Figures 2(e) and 2(f)).  proliferation. We therefore quanti ed the number of Ki67 + cells in the tail and dorsal skin wounds 7 days after treatment with aerosol therapy. ere was no di erence in the positive areas for Ki67 staining between aerosol PBS and aerosol GCB (Figure 3(a), P > 0.05 for both dorsal and tail wounds). In addition, quanti cation of the positive areas for broblasts using immunohistochemistry also showed no di erence between aerosol PBS and aerosol GCB (Figure 3(b), P > 0.05 for both dorsal and tail wounds). ese results demonstrate that aerosol GCB had no e ect on cell proliferation or the number of broblast cells during wound healing.

Aerosol GCB Did Not A ect the 3 Critical Phases of Wound
Healing. Wound repair typically consists of three phases: initial in ammation, followed by proliferation and nally remodeling. To determine whether aerosol GCB a ects wound-induced in ammation, we quanti ed the number of mast cells in the wound tissues using toluidine staining, the number of macrophages using F4/80 antibody staining, and the ratio of M1 to M2 macrophages. M1 macrophages, which inhibit cell proliferation and cause tissue damage, were identi ed and quanti ed using iNOS staining. M2 macrophages, which promote cell proliferation and tissue repair, were identi ed and quanti ed using the anti-mannose receptor antibody. Both types of macrophages are important for normal wound healing as is the M1/M2 ratio. Aerosol GCB had no e ect on the number of mast cells (Figure 4 Figure 3: E ect of aerosol GCB on cell proliferation and broblast numbers associated with wound healing. e dorsal and tail skin wounds from the mice treated with aerosol PBS or aerosol GCB were examined on day 7. (a) Cell proliferation was assessed using Ki67. (b) Fibroblast numbers were assessed using anti-broblast antibody. e positive areas were quanti ed by SimplePCI groups obtained from ve h.p.f areas and compared using Student's t-test. P > 0.05 for all graphs. 6 Sarcoma We next determined whether aerosol GCB a ected endothelial cell proliferation and angiogenesis, both critical functions in the second phase of wound healing.
Endothelial cells play an important role in angiogenesis in wound-healing process.
To explore whether GCB a ects endothelial cell growth and recruitment in wound healing, we quanti ed the number of CD31 + cells (a marker expressed by endothelial cells) in dorsal and tail wound tissues 7 days after aerosol PBS or aerosol GCB.
ere was no di erence in CD31 staining in the tail and dorsal skin wounds between aerosol PBS-and aerosol GCB-treated mice ( Figure 5(a), P > 0.05). VEGFR-2 has been shown to mediate almost all of the known cellular responses to VEGF [29]. We therefore measured the density of VEGFR-2 + cells in the wound areas and once again found no di erence between the aerosol PBSand aerosol GCB-treated groups ( Figure 5(b), P > 0.05). ese results indicate that aerosol GCB did not inhibit angiogenesis (an important process in wound dealing) or interfere with the growth of endothelial cells.
To determine whether aerosol GCB a ected tissue remodeling (the third phase of wound healing), we quanti ed IL-10 and FGF-2 (bFGF), both of which are critical cytokines for the remodeling phase of wound healing.
IL-10 plays an important role in wound healing through its function to inhibit the in ltration of neutrophils and macrophages toward the site of the wound area during this phase [28]. We measured the positive areas of IL-10 in dorsal wound and tail wound sections 7 days after GCB treatment. Again, we showed no di erence in tail or dorsal wounds ( Figure 6(a), P > 0.05) between aerosol GCB-and aerosol PBS-treated mice. FGF-2 or bFGF mediates the formation of new blood vessels that are critical in the wound-healing process following surgery. Di erent studies demonstrated a correlation between reduced FGF-2 expression and wound-healing disorders [16,30,31,33]. mRNA levels of FGF-2 were reduced during wound healing in healing-impaired genetically diabetic mice compared with control mice [38]. Expression of FGF-2 was found to be upregulated after injury in normal but not in diabetic rats. Impaired would healing was seen in aged mice, and this impairment was associated with reduced levels of FGF-2 [33]. Finally, when FGF-2 null mice were used for wound-healing studies, they showed delayed healing, while there was no delay seen in FGF-1-knockout mice [30,31] Similar to our ndings with IL-10, there was no di erence in the expression of FGF-2 (bFGF) in the dorsal and tail wounds between the mice treated with aerosol PBS and aerosol GCB (Figure 6(b), P > 0.05).

E ect of Aerosol GCB on Wound Healing following
Amputation. Healing of the bone and tissue following amputation and tumor removal is a critical part of the patient recovery process. We therefore monitored healing of the amputated area every other day. e amputation woundhealing process in the mice following aerosol GCB administration was not delayed compared with the mice treated with aerosol PBS. Similar to what we observed in the tail and dorsal skin wounds, there was no di erence in Sarcoma broblast proliferation, CD31 expression, VEGFR-2 expression, or the wound-associated cytokines FGF-2 (bFGF) and IL-10 between aerosol PBS and GCB groups (Figure 7).

Discussion
Combination chemotherapy given both pre-and postoperatively has raised the overall survival of patients with primary nonmetastatic osteosarcoma from 20 to 65% [1][2][3]. However, despite the use of aggressive multiagent chemotherapy, 30-35% of patients who have no detectable metastasis at the time of diagnosis develop pulmonary metastases following surgical resection and adjuvant chemotherapy. is statistic has not improved in >30 years, and patients who develop lung metastases have a signi cantly reduced longterm survival [4,5]. Equally disturbing is that there has been relatively little success in treating relapsed patients with surgical resection of metastases in the lung, which is the most e ective approach [32]. A recent analysis showed that the median time to progression in multiple Phase II trials in the Children's Oncology Group for children and adolescents with relapsed osteosarcoma was ∼4 months [34]. Furthermore, no responses were seen in multiple Phase I trials from a single institution [35]. Due to the absence of e ective secondary agents and the poor response rates to date for relapsed patients, it is critical to identify conditions that put the patients at a higher risk for relapse.
Postoperative chemotherapy is usually not initiated for 2-8 weeks after tumor resection, depending upon the postoperative course, as chemotherapy can interfere with the wound-healing process and the production of cytokines at the wound site that have been shown to be critical to the healing process.
is leaves the patient in a potential vulnerable setting as the tumor cells are free to divide and grow unchecked. Delays > 2 weeks postoperatively correlate with a poorer overall survival [7,8]. Two independent retrospective studies showed that increased time from the de nitive surgery to the resumption of chemotherapy was associated with an  Figure 5: E ect of aerosol GCB on CD31 and VEGFR in the wound area. e dorsal and tail skin wounds from the mice treated with aerosol PBS or aerosol GCB were examined on day 7 using (a) anti-CD31 or (b) anti-VEGFR e positive areas were quanti ed by SimplePCI groups obtained from ve h.p.f areas and compared using Student's t-test. P > 0.05 for all graphs.
8 Sarcoma increased risk of death in patients who presented with localized osteosarcoma [7,8]. A study from the Children's Oncology Group (COG) showed that, of 703 patients analyzed, resumption of chemotherapy ranged from 3 to 97 days following resection, with >50% of patients going >2 weeks without treatment and 22% having chemotherapy delays in excess of 21 days [8]. e times to chemotherapy resumption following surgery were dependent on several factors including the type of surgery and surgical margins. Location of the tumor (proximal versus distal extremity), tumor size (< or ≥9 cm), and % necrosis did not play a role. ere was a signi cant decrease in overall survival in those patients that had a delay of >16 days. Patients with a delay in excess of 21 days had a 57% increased risk of death. ese ndings were con rmed in a more recent analysis of 77 patients with a median follow-up time of 11 years [7]. Prolonged time to the resumption of chemotherapy was once again associated with an inferior overall survival, and 75% of patients had delays beyond 21 days. Additionally concerning is that a recent study showed that the median time interval between the last neoadjuvant chemotherapy before surgery and the rst dose of chemotherapy postoperatively was 50-63 days. is prolonged interval with no treatment gives the micrometastases in the lung that were not eradicated by the neoadjuvant therapy the opportunity to grow. Once a patient relapses with lung metastases, e ective therapeutic options are limited as salvage chemotherapy has made no impact on long-term survival [36]. erefore, the time a patient spends in the postoperative period without receiving chemotherapy puts the patient at signi cant risk. Finally, patients with a poor tumor response were found to have a signi cantly increased risk for relapse if there was a delay of >24 days postoperatively in the initiation of chemotherapy compared to those with a good tumor response [9].
is is the most vulnerable population as histologic response correlates with both disease-free and overall survival [9]. As chemotherapy delays increased, both populations were at higher risk for relapse and decreased survival [9]. erefore, identifying treatments that can kill residual tumor cells in the lung which can be initiated in the immediate postoperative period because they have no adverse e ect on the wound-healing process has the potential to decrease the relapse rate in both good and poor responders. Such a strategy can have a signi cant impact on the long-term survival of patients.
Our current investigations show that aerosol GCB results in signi cantly lower serum levels compared with its systemic administration. e dose used in our studies was 1/10th the systemic dose normally given to evaluate e cacy [22]. We have previously evaluated the e cacy of aerosol versus i.p. GCB against both primary OS and OS lung metastases [22]. Treatment in these studies was similar to that in the present investigations. When the primary tumor reached 130 mm 3 and micrometastases were present in the lung, the mice were treated 3 times weekly for 3 weeks. While aerosol GCB was e ective against both the primary tumor and the lung metastases, i.p. GCB at the equivalent dose was only e ective against the primary tumor. GCB given i.p. was not e ective against the OS lung metastases [22]. Since i.p. GCB was not e ective against lung metastases, we did not evaluate the e ect of systemic GCB on wound healing. Our present studies focused on evaluating the simultaneous e cacy of aerosol GCB against pulmonary metastases and its e ect on wound healing. e therapeutically appropriate dose of i.p. GCB would yield even higher serum GCB levels which are more likely to interfere with the healing process. e goal of our study was to evaluate the simultaneous activity of aerosol delivery of the drug on established lung metastases in the setting of a surgical resection. Supportive of the safety and tolerability of aerosol GCB are investigations showing that there was no organ or hematologic toxicity following 5 weeks of aerosol GCB given 3 times per week [21,22]. More importantly, our present investigations show that aerosol GCB initiated 48 hours following amputation of the limb with the primary osteosarcoma tumor inhibited the growth and eradicated established micrometastases in the lung without interfering with wound healing in the skin or bone. Aerosol GCB had no e ect on the 3 phases of wound healing in the dorsal skin, tail, or bone. Cell proliferation, the number of broblasts and mast cells, macrophage recruitment, and the ratio of Type I to Type II macrophages in the dorsal skin and tail wounds of the mice treated with aerosol GCB were similar to those of the mice treated with aerosol PBS.
Anti-VEGF therapy has been shown to result in late wound dehiscence [37]. Aerosol GCB did not inhibit endothelial cell growth or VEGF levels in the wounds and had no e ect on the production of either IL-10 or FGF-2 (bFGF). Similarly, when the wound at the amputation site was examined, treatment with aerosol GCB had no e ect on CD31, VEGF, bFGF, IL-10, or the number of broblasts. is is the rst demonstration that aerosol GCB can have a therapeutic e ect on lung metastases without interfering with wound healing following surgical resection of the primary tumor, a critical process in the treatment regimen for osteosarcoma. Taken together, our data suggest that the initiation of aerosol GCB in the immediate postoperative period has the potential to eradicate lung micrometastases without fear of interfering with the postoperative healing process in the skin or the bone which can delay the resumption of the systemic chemotherapy. Eradication of lung metastases during this period when systemic chemotherapy has been suspended may prevent relapse and result in an increase in the eventfree and overall survival. Patients who develop lung metastases have a 5-year survival rate of only 20% [4,5].

Conclusion
Salvage chemotherapy has made little impact on this survival rate. Aggressive surgical removal of the metastases in combination with salvage chemotherapy only rescues about 40% of patients [6]. As the safety of aerosol GCB has already been demonstrated in adult patients [24], our data support the concept of using aerosol GCB as a bridge after tumor resection and the initiation of clinical trials using aerosol GCB

Conflicts of Interest
e authors declare that there are no con icts of interest regarding the publication of this article.