Angiogenesis is vital for tumor growth but in well-vascularized organs such as the lung its importance is unclear. This situation is complicated by the fact that the lung has two separate circulations, the pulmonary and the systemic bronchial circulation. There are few relevant animal models of non–small cell lung cancer, which can be used to study the lung's complex circulations, and mice, lacking a systemic bronchial circulation cannot be used. We report here a novel orthotopic model of non–small cell lung cancer in rats, where we have studied the separate contributions of each of the two circulations for lung tumor growth. Results show that bronchial artery perfusion, quantified by fluorescent microspheres (206% increase in large tumors) or high-resolution computed tomography scans (276% increase in large tumors), parallels the growth in tumor volume, whereas pulmonary artery perfusion remained unchanged. Ablation of the bronchial artery after the initiation of tumor growth resulted in a decrease in tumor volume over a subsequent course of 4 weeks. These results demonstrate that although the existing pulmonary circulation can supply the metabolic needs for tumor initiation, further growth of the tumor requires angiogenesis from the highly proliferative bronchial circulation. This model may be useful to investigate new therapeutic approaches that target specifically the bronchial circulation. Cancer Res; 76(20); 5962–9. ©2016 AACR.
For tumor growth, new blood vessels are necessary if a tumor is to grow to any substantial size beyond that which can be supplied by simple diffusion from the existing circulation (1). Although this vascular need in tumors has been clearly demonstrated in systemic tissues (2), how it manifests itself in the lung is not entirely clear. The lung is unique because it has two distinct sources of blood; the deoxygenated pulmonary circulation and the systemic bronchial circulation that derives from the aorta. One might imagine that with the pulmonary circulation providing a conduit for the entire cardiac output, there would be more than sufficient blood flow for tumor growth. In fact, in non–small cell lung cancer, there have been some reports of nonangiogenic tumors sustaining themselves by vascular co-option (3, 4). However, there is good evidence that the pulmonary circulation has very limited angiogenic capacity, whereas the bronchial circulation has been shown to have prolific and substantial angiogenic properties (5, 6). Indeed, there are several studies that clearly show a preference for the bronchial circulation by lung tumors (7–9). It may be that the pulmonary vasculature can function as a maintenance vasculature for small tumors but, with little capacity for angiogenesis, the proliferating bronchial vasculature provides the primary source for new vessels and subsequently, tumor growth. The vascular supply of solid lung tumors has not been well studied, because there are few good animal models of lung cancer.
This lack of a good lung cancer model has limited mechanistic studies of the vascular sources, in part due to the difficulty of separating and measuring pulmonary versus bronchial perfusion. Also, because of the difficulty of taking multiple serial measurements tracking tumor size and perfusion, studies in humans with lung cancer have only measured perfusion associated with a single tumor volume, providing no information about the temporal changes to each vascular bed with increasing tumor size. Common xenograft models in rodents using the flank lack the circulatory complexity of the lung, and mouse lung models are inadequate due to the fact that mice lack a subcarinal bronchial vasculature (10). A further complicating factor is that in the process of tumor vascular growth, new anastomoses can form between the bronchial and pulmonary circulations. So even if the bronchial circulation is primary, such anastomoses could possibly allow the pulmonary circulation to supply a secondary nutrient flow to the tumor. Indeed, there is radiographic evidence suggesting that pulmonary anastomoses can supply metastatic lung tumors (11). Each of these methodologic challenges has been an obstacle to mechanistic studies defining the dual circulations involved in lung tumor perfusion.
In this article, we describe a novel rat model of NSCLC and quantify bronchial and pulmonary tumor perfusion independently as they change with tumor growth. We show that impairment of the bronchial circulation cannot only stop tumor growth, but also may lead to diminishing tumor size. This model opens new therapeutic possibilities for studying treatment of primary lung cancer.
Materials and Methods
In vitro determination of angiogenic potential of endothelial cells
Endothelial cell isolation.
Primary endothelial cells were isolated from the pulmonary artery (PAEC) and pulmonary microvasculature (MVEC) using methods described previously (12). The bronchial artery was visualized on the dorsal surface of the trachea and main stem bronchus and dissected. Bronchial artery endothelial cells (BAEC) were isolated following the same protocol. Cells were maintained in 2% FBS cultures under basal conditions and were studied between passage 3 and 5.
FACS analysis of endothelial cells.
To determine proliferation after a 24-hour incubation period with adenocarcinoma supernatant, endothelial cells were stained with thrombomodulin (anti-CD141; Bioss Antibodies) and live cells were selected using VIVID (Invitrogen). For the cell proliferation marker, cells were incubated with fixation/permeabilization buffer (eBiosciences) and intracellularly stained with fluorescence-labeled anti-Ki67, an antigen found in all active phases of the cell cycle (BD Pharmingen). Sample profiles were acquired on a BD FACSaria (BD) and data were analyzed with FlowJo Software (TreeStar).
BAECs, PAECs, and MVECs were cultured in the top chamber of transwell plates (6.5-mm diameter inserts, 5.0-μm pore size polycarbonate transwell membrane; Corning Incorporated) in 2% FBS, whereas adenocarcinoma supernatant was added to the bottom chamber. Cells in the bottom chamber were counted using a hemocytometer after 24 hours of incubation.
Endothelial tube formation.
BAECs, PAECs, and MVECs were grown in a 3D culture (Matrigel, growth factor reduced basement membrane matrix; Corning Incorporated) with adenocarcinoma supernatant for 24 hours. Each endothelial cell type, from nine different animals, was cultured in triplicate wells. Images from three fields of view per well were taken at ×100 original magnification (Olympus IX51 microscope and High Performance SensiCam), and total tube lengths of connecting cells from the three images were measured by use of Image Pro Plus 5.1 software (Media Cybernetics).
In vivo lung adenocarcinoma model
Adenocarcinoma cell delivery.
The experimental protocol was approved by the Johns Hopkins Animal Care and Use Committee (Protocol #RA12M283). Human lung adenocarcinoma cells (Calu-3, ATCC; tested and authenticated by vendor's cytogenetic analysis) were grown in culture, and 5 × 106 cells in 50 μL of sterile saline were injected transthoracically (without thoracotomy) into the left lung of anesthetized, ventilated RNU nude rats (Charles River Laboratories).
Tumor volume by high-resolution computed tomography scan.
Rats were anesthetized with a ketamine/xylazine (75/25 mg/kg) solution. High-resolution computed tomography (HRCT) scans were performed in the Department of Radiology, CT, at the Johns Hopkins Outpatient Center, on a Seimens Definition Flash 256 slice dual source scanner. Data analysis and scanner specifications are described in Supplementary Fig. S1.
Tumor perfusion by fluorescent microsphere injection.
Differential tumor perfusion from the bronchial versus the pulmonary circulation was determined using two different fluorescent microsphere colors. Surgical procedures, tissue digestion, and fluorescence quantification were described previously (13). In addition to the previously described protocol for the bronchial circulation, a femoral vein catheter was infused for quantification of pulmonary perfusion (106, 10 μL yellow polystyrene fluorescent microspheres; Invitrogen). After exsanguination, lung tumors were visualized and carefully dissected from lung tissue for digestion and fluorescence analysis of crimson (bronchial) and yellow (pulmonary) microspheres.
Tumor perfusion by contrast-enhanced HRCT.
Rats were anesthetized and a catheter was placed in a femoral vein for contrast (Visipaque 320; GE Healthcare) infusion. Image analysis and scanner settings can be found in Supplementary Fig. S2.
Bronchial artery ablation.
After an initial measurement of tumor volume by HRCT (4 weeks), rats were anesthetized, ventilated, and a thoracotomy was performed. The left lung was repositioned with forceps and the airway was exposed to visualize bronchial arteries on the dorsal surface of the left main stem bronchus. Bronchial arteries were cauterized using a Gemini Cautery System (Braintree Scientific). In sham animals, the same surgical procedure was followed, but an intercostal artery was cauterized instead of the bronchial artery. The lung was expanded and the thorax closed.
All data are presented as the mean ± SE. A one-way ANOVA with Bonferroni's multiple comparison test was used to evaluate proliferation, chemotaxis, and tube formation. Log-transformed regressions were performed to measure the correlations between tumor weight and number of microspheres as well as tumor volume and voxel intensity. t tests were used to compare perfusion and volume between small and large tumor groups. A P value ≤ 0.05 was accepted as significant.
In vitro determination of angiogenic potential of endothelial cells
BAECs, PAECs, and MVECs were stimulated with adenocarcinoma supernatant to determine the angiogenic potential of each endothelial cell type. When quantifying Ki67+ staining as a measurement of proliferation, basal rates of proliferation varied among cell type and between experiments (2–15% Ki67+) but showed no overall group differences. After normalization to average baseline proliferation for each cell type, exposure to adenocarcinoma supernatant resulted in significantly increased levels of proliferation in BAECs (P < 0.0001) when compared with both PAECs and MVECs (Fig. 1A). Under basal conditions, there was no chemotaxis among any cell type. Yet BAECs showed significantly greater chemotaxis in response to adenocarcinoma supernatant than PAECs (P = 0.04; Fig. 1B). In 3D Matrigel cultures, there was little tube formation in any cell type under basal conditions. However, upon supernatant stimulation, the sum of tube lengths (μm) was significantly greater in BAEC cultures than PAEC and MVEC (P < 0.001; Fig. 1C). To further confirm the angiogenic potential of each endothelial cell type, VEGF protein, one growth factor, known to be secreted by adenocarcinoma cells (14, 15), was used for stimulation in each experimental setting. When endothelial cells were stimulated with VEGF (10 ng/mL), a significant increase in proliferation (83% increase compared with both PAECs and MVECs; n = 9, P = 0.0005) and tube formation (94% and 93% increases compared to PAECs and MVECs, respectively; n = 6, P < 0.0001) were observed. The results of these in vitro assessments of angiogenic potential demonstrated a markedly enhanced responsiveness in isolated BAECs relative to PAECs and MVECs.
In vivo lung adenocarcinoma model
The model we use involves direct injections of adenocarcinoma cells through the chest wall into the left lung. We used different techniques as described below to measure the growth of tumor mass and perfusion as a function of time after the injection. These measurements were also made after ablation of the bronchial artery.
Tumor mass by HRCT.
To confirm the presence of tumor formation after adenocarcinoma cell injection, HRCT scans were performed weekly after the delivery of adenocarcinoma cells directly into the left lung. Representative thoracic HRCT images from a single rat depict four time points after adenocarcinoma cell injection (Fig. 2A). In this specific rat, tumor detection was not possible until 3 weeks after adenocarcinoma cell injection. However, in other rats, tumors were detected as early as 2 weeks. The HRCT scans at 4 and 8 weeks demonstrate the progressive increase in tumor size over time. Tumor volume was calculated from the summation of the serial regions of interest (white outline) on the image slices containing the tumor (see Supplementary Methods). The growth patterns observed in Fig. 2B depict a progressive increase in tumor volume (cm3) over the course of 8 weeks, although the rate of growth varied among animals (A–F; Fig. 2B). To confirm the validity of HRCT to quantify tumor mass, in a separate series of seven rats, tumor volume measured by HRCT was shown to significantly correlate with tumor weight (r2 = 0.98, P < 0.0001; Fig. 2C).
Tumor perfusion by microsphere injection.
Fluorescent microsphere injection was used to quantify tumor perfusion from the pulmonary or bronchial circulations. By using two different colored microspheres and dual injection sights (carotid artery to aorta: representing the bronchial circulation, and femoral vein to pulmonary artery: representing the pulmonary circulation), the contribution of each circulation to total tumor perfusion was quantified. The total number of microspheres counted in the tumor (bronchial + pulmonary perfusion) significantly increased with tumor weight (r2 = 0.541; P = 0.0018; n = 15; Fig. 3A). When tumor weights from all rats were dichotomized to less than or greater than 0.5 g, the larger tumors had a significantly higher percentage of microspheres from the bronchial circulation than the smaller tumors (P = 0.003; Fig. 3B). The pulmonary circulation did not change its contribution to tumor perfusion between small and large tumors.
Tumor perfusion by contrast-enhanced HRCT scanning.
Contrast-enhanced HRCT scans were used to independently measure in vivo tumor perfusion [intensity of contrast medium in Hounsfield units (HU) as a fraction of maximum HU; see Supplementary Methods]. Contrast HRCT scanning confirmed that bronchial perfusion significantly increased with tumor volume (r2 = 0.74; P = 0.0003; n = 11; Fig. 4A). When tumors were dichotomized to less than or greater than 0.3 cm3 in volume, there was significantly greater bronchial perfusion to the larger tumors (P < 0.0001; Fig. 4B). Consistent with previous results, there was no difference in pulmonary perfusion as measured by contrast-enhanced HRCT (Fig. 4C), and when tumors were partitioned by volume (<0.3 vs. >0.3 cm3) there was no significant difference in pulmonary perfusion between small and large tumors (Fig. 4D).
Tumor volume after bronchial artery ablation.
To confirm the importance of bronchial vascular angiogenesis in lung tumor growth, the bronchial artery was cauterized 4 weeks after adenocarcinoma cell injection. HRCT was used to determine initial tumor volume in rats 4 weeks after injection of adenocarcinoma cells, and at final tumor volume at 8 weeks. Rats were divided into two groups; one group with an intact bronchial artery (intact BA; n = 8) and a second group where the bronchial artery was cauterized the day after the initial tumor volume scan (ablated BA; n = 7). Two animals (represented by triangles) were sham controls that followed the same surgical procedures as the “ablated BA” group, except an intercostal artery was cauterized instead of the bronchial artery. Figure 5A shows the individual tumor volumes in rats with an intact bronchial artery and in the group with an ablated BA. There was significantly less change in tumor volume between 4 and 8 weeks in animals after bronchial artery ablation (P = 0.0009; Fig. 5B) and in absolute tumor volume at 8 weeks. The tumor volume was significantly decreased in rats after BA ablation (P = 0.0076; Fig. 5C). Contrast-enhanced HRCT scans were performed at the time of the 8-week tumor volume measurement to confirm the sustained reduction in bronchial tumor perfusion resulting from the bronchial artery cauterization. In animals with BA ablation, there was a significant decrease in both bronchial tumor perfusion and the slope of the relationship between tumor volume and bronchial perfusion compared with intact BA.
Folkman first described the vital role angiogenesis plays in tumor growth and suggested that inhibiting it would lead to tumor necrosis and cancer cell death (2). However, the manifestation of this conjecture has been inconsistent since that time. In particular, in the lung, it is possible for a growing tumor to utilize the existing vasculature by vessel co-option, suggesting neovascularization is not always necessary for tumor establishment and growth (3, 16). However, given the diminished capacity for the pulmonary circulation to undergo angiogenesis (17), a solid tumor might outgrow what can be supplied by the pulmonary circulation, whereupon it would then require its own additional blood supply. The dynamics of this process in conjunction with tumor growth in the lung have not been studied. Several groups have utilized contrast-enhanced HRCT studies in patients to quantify functional tumor perfusion (8, 18, 19). Yuan and colleagues introduced refined CT perfusion protocols in a small study of patients with diverse primary lung carcinomas and showed that tumor angiogenesis was somewhat size-dependent overall (7), but the systemic circulation played the dominant role. However, longitudinal tracking information of tumor volumes with vascularization in individual subjects is not available in patients or animal models. This study was undertaken to define the parallel processes of solid tumor growth in the lung and the corresponding angiogenesis, while simultaneously identifying the specific supporting vasculature.
Essential to this study was a physiologically relevant animal model that would allow for longitudinal measurement of tumor volumes and perfusion sources. We developed a model of NSCLC in nude rats, resulting in the development of a single solid tumor. Substantial tumor growth took place over the course of 8 weeks, as confirmed by tumor weight, tumor volume by HRCT, and lung pathology. Histologic sections showed clear separation between proliferating tumor and normal lung parenchyma, with additional necropsy confirming the absence of any systemic metastases. Once the model was established, the source of tumor perfusion in this rat model could be quantified. It is worth noting that such a study of the lung tumor vascular source cannot be done in mouse models. Mice lack a subcarinal bronchial artery (10, 20), so cannot be used to model the source of tumor perfusion in humans. Consequently, the results of several mouse models including Lewis lung carcinomas (21), those with the overexpression of oncogenes (22–24), or of xenografts in the flank (25, 26) may have limited applicability to the central question of perfusion pathology.
In our study, we provide two functional measurements with complementary information concerning the source and proliferation of the vasculature supplying the growing tumor. Different colors of fluorescent microspheres infused separately into the pulmonary circulation and the systemic bronchial circulation confirmed the ever-increasing systemic perfusion of the growing tumor. In small tumors, perfusion from the bronchial circulation was less than 10% of total tumor perfusion. However, in larger tumors, bronchial perfusion increased significantly to an average 30.6% of total tumor perfusion (average 206% increase). Because these were terminal experiments with only one tumor weight measured at completion of the experiment to pair with microsphere perfusion, the use of HRCT to monitor longitudinal tumor growth provided important additional information. With perfusion HRCT, a similar overall change in perfusion was measured. In small tumors, systemic bronchial perfusion constituted approximately 21% of total perfusion. Beyond the 0.3 cm3 median size of tumors studied, bronchial perfusion increased an average of 276%. Differences in magnitude of the initial perfusion with the two methods likely can be attributed to different tumor sizing methods (weight vs. calculated volume) and different median sizes for the analysis. Nevertheless, the changes in bronchial perfusion with the two different functional measurements were quite similar in the magnitude of increased perfusion (206% vs. 276%). Importantly, both perfusion measurements demonstrated that the pulmonary circulation remained relatively constant and did not increase with tumor size. These functional results demonstrate that the bronchial circulation undergoes angiogenesis to support the needs of the growing lung tumor over the course of 8 weeks, whereas the pulmonary vasculature remains unchanged.
To further confirm the critical role of bronchial angiogenesis in tumor growth, we surgically intervened and studied subsequent tumor growth after the elimination of the bronchial circulation. After 4 weeks of tumor growth, the left bronchial artery was cauterized. After this ablation, there was no further change in tumor volume for at least 4 weeks after the surgical intervention. Final tumor volumes at 8 weeks were significantly smaller than the 8-week tumor volumes in animals with an intact bronchial artery (Fig. 5). Perfusion measurements made at the corresponding 8-week tumor volume determination confirmed the continued absence of any new systemic blood flow, as determined by the significant decrease in the both the baseline and slope of the relationship between bronchial perfusion and tumor size. Thus, angiogenesis of the bronchial circulation is critical for tumor growth. Without new blood vessels supplied by the systemic bronchial vasculature, lung tumors could not grow beyond the minimal size, which could be supported by the pulmonary circulation alone.
The results of bronchial artery ablation should be put in context with studies in humans attempting to modify this vascular source. As early as 1969, bronchial artery infusion therapy was used as a route for more direct delivery of chemotherapy to lung tumors in patients with lung cancer (27), and such approaches to treat NSCLC and hepatocellular carcinoma lung metastases are still being used (28, 29). It is also common for patients with NSCLC to experience life-threatening hemoptysis, and for decades it has been routine practice to embolize small branches of the bronchial artery as an effective treatment (30, 31). In cases where bronchial artery embolization has been performed, there have been no reports on how this procedure specifically alters tumor growth. Because only small branches of the bronchial artery are typically embolized and patients generally are also undergoing chemotherapy, results of embolization per se on lung tumor growth have not been established. It appears that a more complete bronchial artery ablation has yet to be used as a therapeutic option in advanced disease states.
Although the focus of this study was not to identify relevant growth factors required for the process of bronchial vascular angiogenesis, our initial observations regarding bronchial endothelium provide some insights. Bronchial endothelial cells showed robust proliferative characteristics, chemotaxis, and tube formation in response to both adenocarcinoma supernatant and VEGF, compared with pulmonary macro- or microvascular endothelium. As these in vitro measurements are frequently used as exclusive indicators of angiogenesis, the interpretation of many previous studies where examination of relevant endothelial cell types was not considered becomes very difficult. Given the focus of many cancer therapies on tumor angiogenesis (32–34), recognizing and exploiting differences in endothelium seems essential.
Although this study was designed to examine the vascular contributions to a solid NSCLC tumor, it is unknown whether the results might be extrapolated to other lung cancer cell types. Small cell carcinoma or bronchial carcinoma may have different perfusion patterns dependent on tumor size and location. It is reasonable to suggest using this model with the injection of a small cell carcinoma, or xenograft obtained directly from patient samples, to further examine tumor perfusion patterns. Clinically, contrast-enhanced CT scanning could be performed in patients with NSCLC to determine the perfusion contribution from the bronchial circulation, therefore dictating subsequent treatment protocols. Given how effective bronchial artery ablation was in limiting tumor growth in our model, it might be expected to similarly compromise lung tumor growth in patients and other models of lung cancer. Of course, from a potential therapeutic perspective, limiting the total systemic arterial perfusion to an organ may hurt the organ as much as the tumor. So only tumors in the liver with its two circulations would be very relevant to our present model in the lung. Indeed, it is known that the systemic hepatic circulation provides the primary blood flow to liver tumors (28), much like the bronchial circulation being essential for large tumors in the lung parenchyma as shown here.
Tumors have been shown to utilize several methods for vascularization including angiogenesis and vessel co-option. Given the extensive pulmonary capillary network, lung tumors may initially not require extensive angiogenesis. However, if the tumor is to grow more massive, the results of this study demonstrate the essential role for the proliferative capacity of bronchial endothelium compared with the pulmonary vasculature. Tumor growth required bronchial angiogenesis, so targeting the bronchial endothelium may provide opportunities to develop new therapeutic approaches.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: L. Eldridge, R.H. Brown, W. Mitzner, E.M. Wagner
Development of methodology: L. Eldridge, R.H. Brown, W. Mitzner, E.M. Wagner
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L. Eldridge, J. Jenkins, M. Snyder, R.H. Brown, W. Mitzner, E.M. Wagner
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L. Eldridge, M. Snyder, R.H. Brown, W. Mitzner, E.M. Wagner
Writing, review, and/or revision of the manuscript: L. Eldridge, J. Jenkins, M. Snyder, R.H. Brown, W. Mitzner, E.M. Wagner
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Eldridge, A. Moldobaeva, Q. Zhong, J. Jenkins, W. Mitzner
Study supervision: E.M. Wagner
The authors acknowledge NHLBI for supporting this work (HL10342 and HL113392).
This work was supported by NHLBI grants HL10342 and HL113392.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.