INTRODUCTION

It is well accepted that the intratumor microenvironment, such as oxygenation status, greatly influences the radiosensitivity of tumor cells and that the intratumor microenvironment is closely related to the functional status of tumor microvasculature. Therefore, detailed insight into the radiation-induced changes in tumor microvasculature is important for maximizing the efficacy of radiotherapy against cancer. The radiation-induced changes in tumor blood vessels have been demonstrated to be markedly variable, depending on the total radiation dose, dose rate, fraction size and the number of fractions as well as on biological factors such as the tumor type, tumor site and the stage of tumor growth. In the early years of radiotherapy, tumors were irradiated with single or a few fractionated large doses (1, 2). However, after the landmark observations by Regaud and Ferroux (3) and Coutard (4) about 80 years ago that the therapeutic ratio in treating cancer with radiation could be increased by delivering the radiation in multiple fractions of small doses, fractionated radiotherapy has been an almost universally accepted clinical practice. As early as 1936, Mottram (5) reported that the cancer cells in the peripheral tumor volume with good blood supply were more sensitive to radiation than the cells in the central tumor volume. Subsequent studies demonstrated that the response of tumor cells to radiation is closely related to oxygen supply through blood perfusion and that fractionated radiotherapy minimizes radiation-induced vascular damage, thereby allowing reoxygenation of hypoxic tumor cells. In the 1950s, radiosurgery using gamma knives was introduced by Leksell (6) to deliver high-dose hypofractionated radiation to confined vascular lesions and malignancies in the brain. As a result of the recent remarkable advances in imaging technology, computer-aided field shaping, treatment planning, dosimetry and radiation delivery systems, it has now become possible to conformally deliver 20–60 Gy to tumors in a single fraction or 2–5 fractions (7–11). This is referred to as stereotactic radiosurgery (SRS) for the treatment of intracranial lesions and stereotactic body radiation therapy (SBRT) for the treatment of extracranial tumors. A relevant and important question is the fate of tumor blood vessels when tumors are exposed to such high-dose hypofractionated radiation. Another pertinent question is whether the well-established radiobiological principles of conventional fractionated radiotherapy such as reoxygenation of hypoxic cells play any role in the response of tumors to SRS or SBRT. We have briefly discussed these topics in our recent publication (12). Numerous reports have described the radiation-induced vascular changes in tumors and their implications for radiotherapy but have often reached conflicting conclusions. The purpose of the present review is to examine the previous studies on radiation-induced vascular changes in both human and experimental tumors to establish an up-to-date view of the subject with greater relevance to the rapidly growing interest in treating tumors with high-dose per fraction radiotherapy, such as SBRT and SRS.

TUMOR BLOOD VESSELS AND BLOOD FLOW

Solid tumors acquire blood vessels by coopting neighboring vessels in normal tissues and angiogenesis, that is, sprouting or intussusceptive microvascular growth from existing arteries or veins (13, 14). Tumor blood vessels are also formed by vasculogenesis by adding endothelial progenitor and other stem-like cells that derived from the bloodborne and bone marrow-delivered stem cells to the growing tumor vessels (15). As tumors grow and invade normal tissues, the arteries of normal tissues are incorporated into the tumors. Therefore, varying fractions of tumor vascular beds originate from normal tissues. The hastily formed capillary-like tumor microvasculature is comprised of a single layer of endothelial cells often separated by gaps and is devoid of underlying basement membrane or smooth muscle cells (pericytes) and innervations. However, some of the microvasculature of slowly growing human tumors exhibit a thin layer of smooth muscle similar to the capillaries of normal tissues (13). In general, tumor blood vessels are irregular in diameter with rather narrow tubes connected next to dilated and often sinusoid-like structures. Compared to the well-organized web-like network of homogeneous capillaries in normal tissues, the tumor capillaries are sharply bent, tortuous and branched with multiple dead ends (16). Blood perfusion through such disorganized vascular networks tends to be sluggish and often intermittently stationary. However, it should be stressed that the architecture and morphology of tumor microvasculature and the blood perfusion are markedly heterogeneous and variable, depending on various factors such as tumor type, tumor size and site of tumor growth. Even within a tumor, the vascular distribution and blood perfusion are rather heterogeneous. For example, in some tumors, the central volume is well vascularized and well perfused compared to the periphery, while in other types of tumors the opposite is true. The tumor blood vessels are usually more permeable and leaky compared with the blood vessels in the surrounding normal tissues probably because the tumor vasculatures are morphologically immature (17). The tumor vascular permeability is also significantly variable depending on tumor type. For example, the blood-brain barrier in the normal brain tissue is retained in many brain tumors, and thus the blood vessels in brain tumors tend to be less permeable than the blood vessels of other types of tumors. The interstitial fluid pressure (IFP) of certain types of tumors is known to be elevated owing to the high vascular permeability in tumors (14).

RADIATION-INDUCED VASCULAR CHANGES IN TUMORS

We have summarized the 43 representative studies on the radiation-induced changes in tumor vasculature reported in the last 60 years in Table 1. Of the 43 reports (18–60), the first seven (18–24) describe vascular changes in human tumors and the remainder are related to radiation-induced vascular changes in either human tumor xenografts in rodents or transplanted mouse or rat tumors. Various methods, including colpophotography (18), ¹³³Xe clearance (19), Doppler sonography (20), MR imaging (22) and CT imaging (23, 24), have been used to study the radiation-induced vascular changes in human tumors. All the studies with human tumors are concerned with the vascular changes caused by conventional fractionated radiotherapy. Bergsjö (18) studied the vascular changes in cervical tumors caused by radiotherapy with 17 Gy delivered in 10–12 fractions. Colphophotographic examination of the tumor surface indicated that the vascularity on the surface of tumors improved slightly at the end of the therapy. However, it should be realized that the changes in the vasculature on the surface of tumors as determined with colphophotography may not represent the overall vascular changes in tumors. The conclusions of other studies on human tumors (19–24) follow a general trend that blood flow remains unchanged or increases slightly during the early period of fractionated radiotherapy but decreases toward the end of treatments. For example, in the study by Mayr et al. (22), human cervical squamous carcinoma and adenocarcinoma were treated with 40–50 Gy over 4–5 weeks in 5 fractions/week (2 Gy/fraction). Blood flow, as determined by MRI, increased during the first 2 weeks of therapy and decreased thereafter. The authors reported that high cervical tumor perfusion before the treatment and increasing or persistent high perfusion during the early part of the therapy was a favorable sign of the treatment outcome. Pirhonen et al. (20) reported that a decrease of tumor vasculature during the fractionated radiotherapy of advanced cervical carcinoma was associated with a better treatment outcome and that the patients with increased tumor vascularity at the end of treatment needed further treatment. One may assume that, during the course of fractionated radiotherapy, tumor microvasculature beds gradually become nonfunctional as the demands for nutrients, including oxygen, decline due to radiation-induced death of tumor cells.

In the seven studies with human tumor xenografts (25–31), mostly large-dose irradiations were used in either a single dose or several fractions. An angiographic study by Solesvik et al. (25) showed that 35–45% of 5–15-μm-diameter vessels in human melanoma xenografts became nonfunctional within a week after irradiation with 10–15 Gy in a single exposure, as shown in Fig. 1A. In a xenograft of human squamous cell carcinoma of the larynx (27), irradiation with 10 Gy in a single dose caused a slight increase in the functional vascularity soon after irradiation and a significant decrease at 24 h followed by a gradual recovery to the preirradiation level by 11 days after irradiation. The vascular density in human ovarian carcinoma xenografts irradiated with 20 Gy in 5 Gy/week for 4 weeks was about one-half of that in the control xenografts, as shown in Fig. 1B (28). In a recent study with human glioblastoma xenografts grown in the brain of nude mice (31), functional vascular density and blood perfusion markedly decreased and hypoxic regions increased at 17–18 days after irradiation with 15 Gy in a single dose. However, the tumor blood perfusion began to recover from 3 weeks after irradiation and the tumors started to grow, but the recovery of blood perfusion could be suppressed by inhibiting vasculogenesis by preventing the influx of bone marrow-derived cells.

Mouse or rat tumors grown in transparent window chambers have been used to directly observe vascular changes after treatment. Irradiation with 5 Gy increased both vascular density and perfusion during 24–72 h postirradiation in R3230 mammary adenocarcinoma grown in window chambers placed in the back of rats (34). On the other hand, irradiation of mouse adenocarcinoma in window chambers with 20–50 Gy in a single exposure caused progressive narrowing of blood vessels (32). Irradiation of Lewis lung carcinoma of mice grown in window chambers in dorsal skin with 20 Gy in a single exposure led to uniformly complete destruction and hemorrhage of the tumor blood vessels in 2 days (57).

The results and conclusions of studies with mouse and rat tumors reported over the last several decades are quite variable (35–60) (see Table 1). We can attribute such variable observations to the differences in the method to determine the vascular changes, tumor type used, and the site where tumors were transplanted. Nevertheless, the conclusions of the numerous studies may be generalized as follows. After a single exposure to moderately high doses of radiation, e.g. 5–10 Gy, tumor blood flow initially increases, returning to preirradiation levels or slightly below the preirradiation levels in 2–3 days. After irradiation with 10–15 Gy/fraction once or twice, tumor blood flow decreases soon after irradiation, remains reduced for varying lengths of time, e.g. 1 to several days, and occasionally recovers to the control levels. In most cases, after tumors are irradiated with doses higher than 15–20 Gy in a single exposure, tumor blood flow decreases rapidly followed by deterioration of the vasculature as the tumor volume decreases. Some of the representative studies on the radiation-induced vascular changes in rodent tumors are discussed below. Emami et al. (46) reported that the blood flow in rhabdomyosarcomas grown in the scalp of rats declined by 40–50% within 2 h after irradiation with 16.5–60.5 Gy in a single dose, but the blood flow in the tumors treated with 16.5 Gy recovered by 24 h. In a recent study by Chen et al. (60) with mouse prostate tumors grown in the thigh of mice, irradiation with 25 Gy in a single exposure decreased the vascular density to 25% over a 3-week period, thereby increasing chronic and persistent hypoxic regions in the tumors. Whereas most of the studies with rodent tumor models were conducted using tumors grown subcutaneously in either the thigh or back of animals, Johansson et al. (49) studied the radiation-induced vascular damage in glioma grown in rat brains. Irradiation of tumors with 20 Gy in 5 daily fractions of 4 Gy decreased the vascular intensity to 72% of original levels and also decreased the tumor volume to 77% of original size by 5 days after completion of the treatment. Using Doppler sonography, Kim et al. (57) noninvasively determined the blood flow in Lewis lung tumors of mice grown s.c. in the hind limbs after irradiation with 20 Gy once or twice separated by 2 or 4 days. The tumor blood flow decreased significantly by 2 days after 20 Gy irradiation, but it recovered substantially by 4 days after irradiation. Such recovery of blood flow after the initial 20-Gy irradiation could be successfully suppressed by irradiating the tumors again with 20 Gy at 2 days after the initial irradiation. Reirradiation at 4 days after the first irradiation was less effective than that at 2 days after the first irradiation for sustained reduction of tumor blood flow and for suppressing tumor growth. Tsai et al. (56) also used Doppler sonography and immunohistochemistry to determine the blood flow and intratumor microenvironment in murine melanoma tumors irradiated with 12 Gy in a single exposure. The authors reported that there were marked defects in vascular perfusion and decline in tumor vascularity accompanied by prominent regions of hypoxia, necrosis and hemorrhage when the tumor volume increased 10-fold after irradiation.

We have extensively investigated the radiation-induced vascular change in the Walker 256 tumors grown subcutaneously in the hind legs of rats (39–41). In general, irradiation with doses smaller than 2.5 Gy caused a slight decrease in the functional vascular volume for 6–12 h, followed by a return to preirradiation levels. Irradiation with 5–20 Gy in a single exposure decreased the vascular volume in dose- and time-dependent manner. As shown in Fig. 2, the functional intravascular volume in Walker 256 tumors decreased for 2–6 days after irradiation with 5–10 Gy in many but not all tumors (39). However, irradiation with 30 or 60 Gy in a single dose caused marked and lasting decreases in functional vascular volume or vascularity in the tumors. The decrease in the vascular volume by irradiation with doses higher than 10 Gy was statistically significant (P < 0.001). The vascular permeability in the Walker 256 tumors of rats, as assessed by the extravasation of ¹²⁵I-labeled albumin, was 20–30 times greater than that in the muscle (17). This report was probably the first demonstration that tumor blood vessels are highly permeable compared with the blood vessels of normal tissue. The vascular permeability in Walker 256 tumors increased immediately after irradiation throughout the delivered dose range of 2–20 Gy and then returned to preirradiation levels in 2–3 days (39–41). Others have observed similar increases in vascular permeability in tumors or normal tissues after irradiation (23, 24, 26, 41, 43, 55). Figure 3 shows the effects of irradiation with 20 Gy in 1, 4 or 8 daily fractions on functional vascular volume and vascular permeability in Walker 256 tumors (61). Clearly, the rapid drop in the functional vascular volume after 20 Gy irradiation in a single dose was more substantial than that caused by 20 Gy given in 4 fractions (Fig. 3A). Furthermore, when tumors were exposed to 20 Gy in 8 fractions (8 × 2.5 Gy), the vascular volume initially increased slightly and then decreased as the number of fractions increased. Unlike the rapid decline in intravascular volume observed after a single 20-Gy irradiation, the extravasation of plasma protein (vascular permeability) increased significantly at 24 h after irradiation with 20 Gy (Fig. 3B). It was evident that the radiation-induced increase in vascular permeability at 1 day after irradiation was dose-dependent, with the largest and smallest increases occurring by 20 Gy and 2.5 Gy, respectively. However, the vascular permeability in the tumors irradiated with 20 Gy decreased markedly at 2 days after irradiation. Taken together, it may be concluded that irradiation of Walker 256 tumors with doses exceeding about 10 Gy/fraction causes considerable vascular damage. Figure 4A shows that when Walker 256 tumors grown in the legs of rats were irradiated with 30 Gy in a single dose, the functional intravascular volume declined rapidly and remained decreased until 15–16 days after irradiation while the tumors grew continuously for 7–8 days and then regressed (39). Most of the regressed tumors (80%) started to grow again from 15–16 days after irradiation, and the functional intravascular volume also gradually recovered. It should be stressed that the vascular volume decreased much sooner than the tumor began to regress. In a recent investigation by Brown and his associates (31, 62), irradiation of human brain tumor xenografts with 15 Gy or 20 Gy caused profound vascular damage and regression of tumors, but the tumors began to recur 2–3 weeks after irradiation accompanied by vasculogenesis caused by bone marrow-derived CD11b⁺ myelomonocytes. It was further observed that the expression of hypoxia-inducible factor-1 (HIF-1) was upregulated in the irradiated tumors and that HIF-1 enhanced the recruitment of bone marrow-derived cells for vasculogenesis (31). In this context, Dewhirst et al. (63) reported that reoxygenation of hypoxic cancer cells after irradiation increased the level of HIF-1, which then upregulated the level of vascular endothelial cell growth factor (VEGF) and other proangiogenic factors that are known to protect the tumor microvasculature. Effective inhibition of revascularization caused by angiogenesis and vasculogenesis during radiotherapy or after tumors have regressed may enhance the response of tumors to radiotherapy and prevent the recurrence of tumors after treatment.

MECHANISMS OF RADIATION-INDUCED VASCULAR DAMAGE

Radiation-induced changes in blood perfusion, functional intravascular volume and extravasations rates (vascular permeability) are directly related to the functional integrity and viability of vascular endothelial cells. In a previous study with bovine endothelial cells in vitro (64), the radiation survival curve could be characterized with a D₀ of 1.01 Gy, a D_q of 0.65 Gy, and an extrapolation number (n) of 1.9. In a study with endothelial cells of normal tissues (65), the D₀ of the radiation survival curve was found to be in the range of 1.2–2.0 Gy and 1.7–2.7 Gy in vitro and in vivo, respectively (65). The radiation survival curve of umbilical cord vein endothelial cells had a D₀ of about 1.65 Gy with a moderate initial shoulder (66). Note that these studies were concerned with the radiosensitivity of normal tissue endothelial cells and not with the tumor-associated endothelial cells. Park and her associates recently developed a novel method for harvesting endothelial cells from cancer and normal tissues of human breast cancer patients and expanding the cell populations in vitro. Figure 5 shows the in vitro radiation survival curves of endothelial cells derived from tumor and normal tissues obtained from two different breast cancer patients (unpublished observations). It is clearly demonstrated that the endothelial cells from breast cancer tissue were significantly more radiosensitive than the endothelial cells from normal breast tissue. Relevant to this conclusion, a recent study by Grabham et al. (67) using human vessel models demonstrated that developing vessels are more radiosensitive than mature vessels.

The death of endothelial cells as a result of direct radiation damage in irradiated tumors would cause focal microscopic or macroscopic vascular damage and eventual malfunction and collapse of the affected capillary-like vessels. As noted above, vascular permeability in tumors increases rapidly after irradiation, probably due to damage in the endothelial cells followed by widening of the gaps between endothelial cells (Fig. 3B, Fig. 4C) (23, 24, 39, 41, 43, 48, 55, 68). The increase in extravasation of plasma due to the increase in vascular permeability may increase the erythrocyte concentration within the narrow capillaries, thereby leading to retardation or stasis of blood perfusion. In addition, the increased permeability of capillaries may increase the extravascular or interstitial plasma protein concentrations, thereby elevating interstitial fluid pressure. The elevation of interstitial fluid pressure above the intravascular blood pressure will cause vascular collapse. Therefore, it is probable that the early decline in functional vascularity after irradiation in tumors may be caused at least in part by collapse of blood vessels as a result of elevation of interstitial fluid pressure. When tumor volume shrinks due to death of parenchymal cells after irradiation, the tumor vascular beds may become further disorganized, aggregated, condensed and fragmented (43). Figure 3 shows that the extent of vascular damage in tumors treated with fractionated radiation was less than that caused by high-dose single fractions, in accordance with the reports by others (55, 60). It is likely that sublethal radiation damage in endothelial cells is repaired during fractionated irradiation and thus the functional integrity of tumor vessels is less impaired.

It is noteworthy that in most of the previous studies on the effects of radiation on tumor vascular functions using rodent tumors or human tumor xenografts, the tumors and varying volumes of the surrounding normal tissues were irradiated simultaneously. Since tumor vascular beds are connected to the vascular networks of normal tissues, it is likely that the vascular damage in the adjacent normal tissues significantly influenced the tumor blood perfusion in the previous studies. Therefore, the inconsistent results on the radiation-induced vascular changes observed in the previous studies with experimental tumors may be attributed in part to the differences in the type and site of tumors studied and also the differences in the volume of normal tissues irradiated. Kioi et al. (31) reported that irradiation of human U251 glioblastoma xenografts growing in the brain and in the back of nude mice reduced blood flow to 10% and 30% of the original value, respectively. In this respect, it is known that tumors growth is significantly retarded when tumors are transplanted into previously irradiated tissues rather than unirradiated tissues, which is commonly known as the tumor bed effect (69). It is believed that angiogenesis in tumors, which originates from existing normal tissue blood vessels, is retarded due to radiation-induced vascular damages in the surrounding normal tissues, and thus the supply of oxygen and other nutrients essential for the growth of tumors is limited. It remains to be investigated whether the vascular changes in tumors treated with conformal irradiation such as SBRT or SRS significantly differ from those in the tumors treated with adjacent normal tissues.

VASCULAR CHANGES AND OXYGEN TENSION IN IRRADIATED TUMORS

The intratumor oxygen tension is controlled by the oxygen supply through blood perfusion and the oxygen consumption rate mainly by the tumor cells. Therefore, the radiation-induced vascular changes may affect the tumor oxygenation. Surprisingly, however, there have been only a few studies that simultaneously measured the radiation-induced changes in tumor microvasculature and the intratumor oxygen tension. In the 1960–1970, Carter and Silver (70), Evans and Naylor (71), Kolstad (72), Bergsjo and Evans (73), and Badib and Webster (74) pioneered investigations of the effects of radiotherapy on the oxygen tension in various human tumors. Unfortunately, the results of these early studies are rather inconsistent and difficult to interpret because the studies were conducted with equipment and methods with limited accuracy and reliability (75). For example, Badib and Webster (74) reported that radiotherapy increased tumor oxygenation, but in this study, the tumor pO₂ was measured at only a single point in each tumor. In recent years, using more advanced and reliable methods, investigators determined the changes in pO₂ in various human tumors caused by conventional fractionated radiotherapy. Dunst et al. (76) determined the pO₂ in human cervical cancer treated with fractionated radiotherapy and reported that the median tumor pO₂ increased significantly when the total dose reached 20 Gy, particularly in tumors that had low baseline pO₂ values. However, the tumor pO₂ declined at the end of treatment, and this appeared to be due to vascular damage. Cooper et al. (77) also reported that fractionated radiotherapy increased median pO₂ in human cervical cancer. On the other hand, Lyng et al. (21, 78) and Fyles et al. (79) observed no significant changes in pO₂ in cervical cancer, and Brizel et al. (80) also reported no changes in pO₂ in head-neck tumors during the course of conventional fractionated radiotherapy. Interestingly, in the study by Lyng et al. (21), little changes occurred in pO₂, while there was a clear evidence of vascular damage in the cervical cancer treated with fractionated radiation. In a well-designed study by Stadler et al. (81), the pO₂ in head-neck tumors decreased significantly by the end of a first course of split-course radiotherapy with 30 Gy, recovered during a 2-week break, and then decreased again by the end of the second course of treatment with 40 Gy. These studies showed that fractionated radiotherapy of human tumors may increase, cause no significant change, or decrease in tumor oxygen tension. As pointed out by Molls et al. (75), the only trend observed in studies was that the pO₂ in human tumors decreased by the end of the course of fractionated radiotherapy. It is entirely unclear why the direction and magnitude of changes in tumor pO₂ are so inconsistent among different studies and even in the same tumor types, e.g. cervical cancer (76–79) and head-neck cancer (80, 81). Tumor size, oxygen measurement technique, pO₂ level before treatment, radiation dose and different time-dose schedules are some of the many factors that may control the direction of changes in tumor pO₂. Importantly, unlike the changes in tumor pO₂ during treatment, the tumor pO₂ prior to fractionated radiotherapy has been shown to be related to the outcome of the treatment. The human cervical tumors with high pO₂ before receiving fractionated radiotherapy responded better than those with low pO₂ to the treatments (82).

Radiation-induced changes in pO₂ in human tumor xenografts or animal tumors have also been investigated. Brurberg (29) studied possible relationships between vascular changes and pO₂ in human melanoma xenografts in nude mice. Irradiation with 10 Gy in a single dose caused no changes in pO₂ in the xenografts in 72 h, while the irradiation reduced the blood perfusion by as much as 40%. In the study by Ceelen et al. (58), irradiation of rat colorectal tumors grown in the hind legs of rats with 5 × 5 Gy significantly reduced the microvascular density but slightly increased the intratumor pO₂. Zywietz et al. (48) treated rhabdomyosarcoma in rats with a total dose of 60 Gy in 20 fractions over 4 weeks and observed that tumor pO₂ increased slightly in the early phase of the treatment but declined as the treatment progressed. The investigators attributed the decrease in tumor pO₂ at the end of treatment to damage in the tumor capillary endothelial cells. The pO₂ in mouse adenocarcinoma decreased significantly in 6 h after 20 Gy irradiation in a single exposure, recovered to control levels by 48 h, and then gradually declined (83). Vaupel et al. (84) reported that irradiation of mouse mammary adenocarcinoma with a single dose of 60 Gy markedly increased tumor pO₂ at 72–74 h after exposure. However, Endrich and Vaupel (85) later suggested that a single large dose of radiation would destroy the tumor microvasculature and lead to parenchymal cell death. In a study by Koutcher et al. (86), irradiation of mouse mammary carcinoma with single doses of 32 or 65 Gy significantly increased the mean tumor pO₂ and reduced the frequency of pO₂ values lower than 2.5 mmHg at 3–4 days after radiation exposure. Unfortunately, in those two studies (84, 86), the tumor pO₂ was measured within 3–4 days after irradiation, where as tumor pO₂ may decline later as the radiation-induced vascular damage becomes significant. In this regard, Goda et al. (83) reported that tumor pO₂ underwent dynamic changes after irradiation with 10, 20 and 40 Gy in a mouse tumor model, and they concluded that repeated monitoring is necessary to know the precise changes in tumor oxygenation in irradiated tumors. Nevertheless, it is rather curious that the tumor pO₂ increased after irradiation with 60 Gy or 65 Gy in view of the possibility that irradiation with such large doses would cause severe damage in the tumor microvasculature, as discussed in the previous section. One may speculate that, in the tumors irradiated with 50–60 Gy, the oxygen demand in tumors is drastically diminished due to rapid death of tumor cells or severe damage to tumor cells that would reduce oxygen consumption before vascular damage is fully expressed (12). Another conceivable explanation is that a single large dose of radiation causes a transient vascular normalization by preferentially destroying the most immature and abnormal portions of the vascular bed, allowing for a reorganization of perfusion through the remaining functional, more mature vasculature. However, it is highly likely that an increase in tumor pO₂ after irradiation with doses as high as 60 Gy is a transitional phenomenon because marked increases in the hypoxic areas could be observed in the immunohistochemical preparations of human tumor xenografts 2–3 weeks after irradiation with 15 Gy or 20 Gy (31, 62) or in mouse prostate tumors after irradiation with 25 Gy in a single dose (60). Likewise, necrotic and hypoxic areas increased significantly in human squamous cell carcinoma xenografts after irradiation with 10 Gy (27) and in mouse melanoma irradiated with 12 Gy (56). Fractions of hypoxic cells in rodent tumors have been demonstrated to be reoxygenated after an exposure to doses as high as 10–20 Gy. It should be noted that the reoxygenation of hypoxic cells refers to an improvement of oxygenation status of hypoxic cells that survived the initial high-dose irradiation, and it does not necessarily indicate that the overall oxygenation status in the irradiated tumors is increased. Furthermore, it does not indicate the extent of cell death including hypoxic cells after the initial high-dose irradiation (45). To our knowledge, the oxygen tension in human tumors treated with high-dose hypofractionated SBRT or SRS has not been investigated.

ROLE OF VASCULAR DAMAGE IN THE RESPONSE OF TUMORS TO RADIOTHERAPY

There have been considerable discussions in the radiotherapy community as to whether the primary effect of ionizing radiation in destroying tumors is directly killing cancer cells or indirectly killing cancer cells via vascular damage. Cramer (87) reported as early as 1932 that interference with tumor blood flow caused by radiation damage to tumor stroma played an important role in the overall response of tumors to radiation. Denis et al. (52) reported that the radiosensitivity of rat mammary tumors correlated with early vessel changes. In support of the notion that the major target of radiotherapy is tumor endothelial cells or vasculature and not tumor parenchymal cells, investigators reported that irradiation caused rapid apoptosis in tumor endothelial cells by promoting acidic sphingomyelinase (ASMase)-mediated generation of ceramide, a proapoptotic second messenger (54, 88, 89). Garcia-Barros et al. (54) concluded that ceramide-mediated apoptosis in tumor endothelial cells leads to secondary death in tumor cells and that radiation-induced endothelial cell death is thus the major player in the response of tumors to radiation at the clinically relevant dose range. Fuks and Kolesnick (90) reported that irradiation of tumors with doses higher than 8–10 Gy in a single exposure causes ceramide-mediated apoptosis in endothelial cells, thereby causing indirect death of parenchymal cells, whereas fractionated irradiation with 1.3–3.0 Gy per fraction induces apoptosis in endothelial cells through other signaling pathways. However, above contention that radiosensitivity of endothelial cells dictates the response of tumor to radiotherapy has been strongly rebuffed by other investigators (91, 92). Budach and his coinvestigators in Suit's laboratory (93) previously reported that the TCD₅₀ (radiation dose that cures 50% of tumors treated) for human or murine tumors transplanted into two different strains of mice with different radiosensitivities was dependent on the radiosensitivity of the tumor cells and not the radiosensitivity of the host stromal cells. To further assess the relationship between the intrinsic tumor cell radiosensitivity and tumor response, Gerweck et al. (94) transplanted radiosensitive DNA-PKcs^–/– and radioresistant DNA-PKcs^+/+ tumor cells into the same strain of nude mice and studied the radiation-induced tumor growth delay. The growth delay of the tumors derived from DNA-PKcs^–/– cells were significantly longer than that of the tumors derived from radioresistant DNA-PKcs^+/+ cells, indicating that the radiosensitivity of tumor cells, not that of stromal cells, dictates the response of tumors to radiotherapy. In a subsequent study by the same group of investigators (59), DNA-PKcs^–/– and DNA-PKcs^+/+ tumor cells were transplanted into nude mice and radiosensitive SCID mice, and the resultant tumors were irradiated with 15 Gy in a single exposure. Whereas the irradiation reduced the functional vascularity only modestly in the tumors induced in the nude mice, the irradiation caused considerable reductions in the number of functional vessels in the tumors grown in SCID mice regardless of the intrinsic radiosensitivity of the transplanted tumor cells. An analysis of the radiation-induced growth delay of the tumors indicated that whereas direct killing of tumor cells was the major determinant of tumor response in nude mice, both direct killing and indirect killing of tumor cells as a result of vascular damage contributed to tumor response in SCID mice. It may be concluded that the contribution of radiation-induced vascular damage to the response of tumors to radiation will be significant only when tumors are irradiated with doses high enough to cause substantial vascular damages in the tumors. Therefore, it is likely that the radiosensitivity of endothelial cells is relatively insignificant in the conventional fractionated radiotherapy using 1.5–2.0 Gy/fraction.

Denekamp (95) estimated that one endothelial cell subtends a segment of a tumor volume containing as many as 2000 tumor cells (Fig. 6). Given that blood vessels are serial tissues, sectional damage in a vessel may induce cessation of blood perfusion throughout the affected vessel. Unless the blood circulation through the affected vessel is reestablished soon, severe deprivation of oxygen and nutritional supply will develop along the damaged vessel leading to avalanche of tumor cells death. Clement et al. (45, 96) reported some time ago that irradiation of rodent tumors with 20 Gy in a single exposure caused marked vascular damages leading to massive killing of tumor cells.

In recent years, increasing numbers of cancer patients have been treated with SBRT or SRS, which delivers 20–60 Gy of radiation in 1–5 fractions (7–11). It would be quite reasonable to expect that in human tumors, like in animal tumors, irradiation with high doses in a single or several fractions over a short period will cause severe vascular damage and make the intratumor microenvironment hypoxic, acidic and nutritionally deprived, thereby inducing indirect tumor cell death. Kirkpatrick et al. (97) and Kocher et al. (98) suggested that the total cell death in tumors receiving high-dose hypofractionated radiotherapy is the product of the direct cytotoxicity of radiation to tumor cells and the indirect tumor cell death caused by radiation-induced vascular damage. An important aspect of the cell death due to vascular damage is that, unlike the direct death, the indirect death caused by vascular damage can occur regardless of the oxygenation status prior to radiation exposure. Fowler et al. (99) reported that 3 fractions of 23 Gy (69 Gy) will be necessary to achieve a good chance of eliminating malignant cells in 1–10 g of tumors, assuming that 10% of the tumor cells are hypoxic. Brown et al. (100) concluded that irradiation with 60 Gy in 3 fractions will reduce the survival of tumor cells by 7.7 logs when 20% of the tumor cells are assumed to be hypoxic, and thus 60 Gy irradiation in 3 fractions would not be able to eradicate all clonogenic cells including hypoxic cells in 1–2-cm-diameter tumors. However, high local control of small tumors has often been achieved in clinical trials for SBRT/SRS with doses apparently insufficient to directly kill all the tumor cells, including hypoxic cells (7–11). Based on computer simulation, which was fitted to 90 sets of clinical data, Kocher et al. (98) concluded that the therapeutic effect of a single radiosurgery in malignant brain tumors cannot be explained without the consideration of vascular effects. The hypothetical cell death mechanism in two tumor types with different endothelial cell radiosensitivity is illustrated in Fig. 7. It is assumed that 10% of clonogenic tumor cells are radiobiologically hypoxic in both tumor types. It is shown that, as the radiation dose is increased, fully oxygenated tumor cells are killed initially and then some hypoxic cells are killed as denoted by “a” and “b”, respectively. With the further increase in radiation doe to about 12 Gy, vascular damage is evoked in the tumors with radiosensitive endothelial cells. Consequently, hypoxic tumor cells are indirectly killed, as denoted by “c”. In the tumors with radioresistant endothelial cells, vascular damage and resultant indirect death of hypoxic cells begin to take place when the radiation dose is increased to about 17 Gy, as denoted by “d”. Based on the results of the numerous preclinical and clinical studies, it may be reasonable to suggest that the threshold radiation dose in a single exposure for indirect death of tumor cells in most of the human tumors may fall in the range of 10–15 Gy.

It should be note that the indirect death of tumor cells as a result of vascular damage may not be the only mechanism that accounts for the response of human tumors to SBRT or SRS. It has been suggested that local high-dose irradiation evokes immune reactions and thereby eradicates the tumor cells that escaped the radiation-induced death (101). In support of such notion, a recent report showed that ablative radiotherapy dramatically increased T-cell priming in graining lymphoid tissues, leading to reduction/eradication of the primary tumor or distant metastasis in a CD8⁺ T-cell-dependent fashion (102). A possible role of cancer stem cells in the response of tumors to SBRT or SRS has also been suggested. The frequent failure of radiotherapy of tumors due to recurrence has been suggested to be caused by cancer stem cells that are able to survive conventional fractionated radiotherapy. Cancer stem cells have been reported to reside in the perivascular niche in tumors (103). It is probable that high-dose irradiation disrupts the niche for the survival of cancer stem cells leading to eradication of radioresistant cancer stem cells.

It is evident that further investigation is warranted to obtain better insights into the vascular damage caused in tumors by high-dose hypofractionated irradiation and the implications of such vascular damage for the response of human tumors to SBRT and SRS. An important question is how the vascular damage caused by high-dose hypofractionated irradiation and the ensuing deterioration of intratumor microenvironment such as the hypoxic, acidic and nutritionally deprived environment would affect the well-established radiobiological principles such as the 4Rs (reoxygenation, repair, repopulation and redistribution) and the linear-quadratic equation.

SUMMARY

An analysis of the studies reported over the last several decades indicates that the functional vascularity in human tumors remains unchanged or improves slightly during the early period of conventional fractionated radiotherapy with 1.5–2.0-Gy daily doses but gradually diminishes during the later part of treatment. Numerous studies with experimental tumors indicated that irradiation with doses higher than 10 Gy in a single fraction or 20–60 Gy in limited numbers of fractions causes severe vascular damage leading to the deterioration of the intratumor microenvironment and indirect death of tumor cells. It is highly likely that similar vascular damage would occur in human tumors irradiated with high-dose hypofractionated radiation. We conclude that the radiation-induced vascular damage and the resulting indirect death of tumor cells play important roles in the response of tumors to high-dose hypofractionated SBRT and SRS. In addition, enhanced immune reactions and increased eradiation of cancer stem cells might be involved in the response of tumors to SBRT or SRS. Further studies to gain better insights into the effects of high-dose hypofractionated irradiation on tumor vasculature are warranted. In addition, whether the 4Rs and the linear-quadratic equation are applicable for SBRT or SRS remains to be investigated.