Cancer Therapy Vol 3, 105-130, 2005

1Department of Radiation Medicine and

__________________________________________________________________________________

*Correspondence: Daila S. Gridley, Ph.D., Chan Shun Pavilion, Room A-1010, 11175 Campus Street, Loma Linda, CA 92354; Telephone: (909) 558-8361; Fax: (909) 558-0825; E-mail: dgridley@dominion.llumc.edu

Key words: radiosensitivity, radioresistance, radio-adaptation, dose-rate effect, inverse dose-rate effect

Abbreviations: atomic bomb, (A-bomb); Central nervous system, (CNS); chronic myelogenous leukemia, (CML); computed tomography, (CT); different poly(ADP-ribose) polymerase-1, (PARP-1); epidermal growth factor receptors, (EGFR); Epstein-Barr virus, (EBV); Food and Drug Administration, (FDA); gastrointestinal tract, (GI); human lung fibroblasts, (HFL-1); hypoxanthine-guanine phosphoribosyl transferase, (HPRT); Lewis lung carcinoma, (LLC); intensity-modulated radiation therapy, (IMRT); linear energy transfer, (LET); lipopolysaccharide, (LPS); magnetic resonance imaging, (MRI); major histocompatibility complex, (MHC); monoclonal antibodies, (Mab); Natural killer, (NK); positron emission tomography, (PET); retinoblastoma, (Rb); subcutaneous, (s.c.); T cytotoxic, (Tc); T helper, (Th); thymidine kinase, (tk); total-body irradiation, (TBI); vascular endothelial growth factor, (VEGF)

In spite of numerous improvements in cancer management, there is an urgent need to implement new treatment approaches. Estimates made by the American Cancer Society indicate that there were 1,368,030 newly diagnosed cases and 563,700 cancer-related deaths in 2004 (Jemal et al, 2004). With current therapies almost 50% of patients will eventually die because of local recurrence, metastatic disease, or a combination of both (Lichter 2000). Furthermore, the rapid expansion of elderly individuals in the United States, as well as in many other countries, will undoubtedly lead to a substantial increase in the prevalence of cancer.

Since the discovery of x-rays by Wilhelm Roentgen more than 110 years ago, radiation has become an important and often indispensable modality for cancer treatment. The development of improved radiotherapy protocols has been closely allied with understanding of factors that determine treatment outcome. One of these has been improved definition of tumors in three-dimensions with the development of magnetic resonance imaging (MRI), computed tomography (CT), positron emission tomography (PET) and molecular imaging. Another improvement has been in the understanding of the physical and chemical processes associated with predicting dose and microdose in tumors and normal tissues. Finally, there has been a better understanding of the radiobiology associated with the response of tumor and normal tissues to different temporal patterns of dose delivery. These areas of improvement could be combined into highly sophisticated radiation therapy protocols using more advanced planning systems. For example, planning systems originally proposed by Slater and colleagues in 1974 have provided guidance in newer treatment modalities such as the use of charged heavy particle and intensity-modulated radiation therapy (IMRT) (Slater et al, 1988, 1992; Purdy, 1996; Mackie and Smith, 1999; Bastasch et al, 2002; Baumert et al, 2003; Teh et al, 2003). All of these advances have contributed to more accurate tumor targeting and thus also to increased sparing of normal tissues. However, understanding is still lacking as to how this improved dose distribution can be used to optimize the therapeutic ratio (tumor cure/normal tissue toxicity). Better understanding of dose-rate effects may aid in such optimization and thus also decrease the need to sometimes underdose the tumor in order to prevent severe normal tissue complications.

It is important for this review to use terminology that is internally consistent and to conform, as far as possible, with conventions in the literature. For this review, we use six dose-rate groups and abbreviations to describe dose-rate ranges: ultra-high dose rate, UHDR (~10⁵ Gy/hr); high-dose rate, HDR (10-100 Gy/hr); intermediate-dose rate, IDR (1.0-10 Gy/hr); low-dose rate, LDR (0.1-1.0 Gy/hr); very-low dose rate, VLDR (0.01-0.1 Gy/hr); and ultra-low dose rate, ULDR (<0.01 Gy/hr). These groups, together with comments are summarized in Table 1. It is also important to point out that while we use these groupings in this review, the dose-rate is not an absolute constant for any radiation because the microdosimetry that describes the relative number of ionizations produced within a microvolume varies even with so called minimally-ionizing radiation such as x-rays or g -rays. The time pattern for delivering radiotherapy has always been recognized as an important variable in outcome. Use of five daily fractions per week (9-10 Gy total/week) has emerged as the standard time-dose regimen for treating many solid tumors. Hyperfractionation and now also hypofractionation, has been evaluated in limited studies in which the fraction size, time between radiation fractions and the total dose are either reduced or increased. However, these protocols still use HDR during radiation delivery. While we believe such experiments may involve some of the same mechanisms that are involved in dose-rate effects, we will not emphasize these types of experiments in this review.

Cellular studies of those dose rates that are used most often in the clinic, ranging from ~0.1 Gy/hr to several Gy/min, have been largely derived from experiments with cultured cells (Hall and Brenner, 1991). The higher dose rates are within the range that we refer to as the HDR range, whereas the lower rates used in brachytherapy (i.e. implantating a radioactive source) extend into the IDR and LDR ranges. Thus, we suggest that the mechanisms that influence cell killing over these different ranges may differ and need to be considered in a more detailed, segmented analysis. Overall, the regimens employed for radiotherapy have been based on empirical observation of outcome, equipment limitations and patient and staff convenience.

Proton radiotherapy has several characteristics that relate to dose and dose-rate considerations. These characteristics are: 1) protons are administered as a rapid sequence of "spills" or small packets of protons and thus

Table 1. Assignment of dose-rates into groups based on temporal chemical and biological processes

the dose delivered is not continual but discontinuous over a very small time scale: While the "average dose-rate" at which protons are administered are comparable to photon high dose-rate (circa 50 Gy/hr) irradiation, instantaneous dose-rates during a "spill" can reach 600 Gy/hr; 2) protons produce a higher proportion of ionization clusters so that for some cell/tissue volumes, there is an increase in effective dose-rate; and 3) improved spatial dose-distribution of deposited radiation with protons permits using larger fractions with lower risk for toxicity compared to photons, a characteristic that we suggest can be used to exploit cell sensitization by low-dose-rate irradiation. Proton therapy is maturing as a first-line radiotherapy for a large number of patients. The clearest example has occurred at the proton treatment facility at the Loma Linda University Medical Center (Slater et al, 1988, 1990; Archambeau et al, 1992). Proton beams, consisting of charged particles that deposit the bulk of their energy as they approach their stopping edge (Bragg peak), are modulated such that the Bragg peak is distributed throughout the intended treatment volume. This then distributes microregions of higher dose-rate through the tumor mass. Because of the lower integral dose delivered during proton radiotherapy, the risk for normal tissue damage is minimized, as evidenced by data presented in rapidly accumulating publications (Matsuzaki et al, 1994; Gridley et al, 1996, 1998, 2004; McAllister et al, 1997; Yonemoto et al, 1997; Slater et al, 1998, 2004; Bush et al, 1999; Rossi 1999; Lin et al, 2000; Schulte et al, 2000; Bonnet et al, 2001; Hug et al, 2002; Kirsch and Tarbell, 2004).

Early studies by Elkind and Sutton utilized split-dose radiation (i.e. fractionation) to simulate low-dose rates (Elkind and Sutton, 1959). Their findings led to the theoretical prediction that biological response to LDR radiation would be decreased compared to the same total dose delivered acutely and that the difference in the effect depended upon the repair efficiency of the resulting damage (Lajtha and Oliver, 1961). The existence of this dose rate sparing was subsequently supported by studies of HeLa cells that were also shown to accumulate in the G2 phase of the cell cycle during LDR exposure (Hall and Bedford, 1964; Bedford and Mitchell, 1973; Mitchell and Bedford, 1977). While the molecular mechanisms that underlie split dose recovery after acute irradiation are unknown, it is important to consider whether this phenomenon relates directly to dose-rate effects.

Later studies, however, demonstrated greater HeLa cell death when radiation was delivered at an LDR rate of 0.37 Gy/hr instead of at 1.54 Gy/hr (Mitchell et al, 1979). This phenomenon became known as the inverse dose-rate effect. The explanation put forth for the increased sensitivity was that cells irradiated at LDR over an extended period of time do not repair all damage; with increasing accumulation in G2, the most radiosensitive phase of the cell cycle, enhanced radiosensitivity is manifested (Mitchell et al, 1979). Indeed, it was long assumed that accumulation in G2 was the major mechanism by which protracted LDR radiation sensitizes tumor cells to HDR radiation. Many reports utilizing a variety of neoplastic cell types have supported this possibility (Terasima and Tolmach, 1961; Bedford and Hall, 1963; Kal et al, 1975; Mitchell et al, 1979; van Oostrum et al, 1990; Knox et al, 1993; Skladowski et al, 1993; Fowler, 2003). The increased sensitivity in G2 may be the result of insufficient DNA repair before cells reach this stage (Parshad et al, 1984) and/or incomplete repair in G2 (Walters et al, 1974; Bases et al, 1980). However, other studies have found no association between G2 arrest and radiosensitivity (Cao et al, 1983; DeWeese et al, 1998; Mitchell et al, 2002). Most recently, it has been demonstrated that exposure to LDR radiation reduces activation of the DNA damage sensor ATM and its downstream target H2AX, resulting in decreased clonogenic survival compared to HDR radiation (Collis et al, 2004).

Most data on dose-rate effects are presented as surviving fraction versus total dose. However the duration of exposure can vary from a few minutes to days. It is important, we suggest, to consider the variation of the rate of cell kill per unit dose as a function of duration of exposure. We illustrate this by replotting data previously reported by Marin, Dillehay and colleagues (Marin et al, 1991) utilizing human U251 glioblastoma cells. These authors observed little dose-rate effects from 0.086 Gy/hr to 49.0 Gy/hr, with survival clustering around the survival pattern for acute exposure. However when these data are plotted as the rate of cell kill per unit time, a different pattern is observed as shown in Figure 1.

Figure 1. Survival curves for human U251 glioblastoma cells exposed to ⁵⁷Cs at variable dose rates. Rate of cell killing expressed as log surviving fraction per Gy as a function of exposure time for dose rates in the very low dose-rate range (0.086 Gy/hr) and in the low dose-rate range (0.123, 0.25, and 0.49 Gy/hr) compared to the rate of cell kill for high dose-rate irradiation (~50 Gy/hr). Data were extracted from the work of Marin et al, 1991 and calculated by measuring the change in surviving fraction in logs for the preceding radiation interval and dividing it by the absorbed dose over the same time interval. The horizontal dotted lines are the rates of cell kill by high dose-rate radiation (~50 Gy/hr) measured as the slope of the survival curve at ~2 Gy (alpha) and for doses higher than 4 Gy (omega). U251 is a very radioresistant cell line derived from a human glioblastoma.

In this figure we compare the rate of cell kill for these several low dose-rates to the rate of cell kill for radiation delivered acutely expressed as the slope at doses circa 2 Gy and at doses higher than 4 Gy. Alpha is the rate of cell kill induced per unit dose by high dose-rate (~50 Gy/hr) at approximately 2 Gy. Alpha is identical to the coefficient alpha as determined by linear-quadratic analysis of the acute survival curve. Omega is related to the inverse of the Do and is calculated by reiterative best-fit analysis of the terminal slope (> 4 Gy) of the acute survival curve. Thus omega represents the maximum rate of cell kill per Gy for acute irradiation. These data when plotted in a time-based format illustrate three important points. First, the rate of cell kill changes dramatically over time for different dose-rates. Second, for this cell line, the rate of cell killing increases as dose rate increases from 0.086 cGy/hr in the VLDR range and reaches a maximum at approximately 0.25 Gy/hr. Increasing the dose-rate to 0.49 cGy/hr does not increase further the rate of cell killing. The third point is most important for considering dose rates other than the HDR used in the clinic, i.e. the data in Figure 1 show that for some time points, the rate of cell kill is equally effective or more effective when delivered at 0.086 to 0.49 cGy/hr, than at rates usually associated with conventional cancer radiotherapy. A fourth point must be made to place these observations into a clinical context. There is a dramatic difference in cell-specific variation between human tumor cell lines observed by Joiner and his colleagues (Marples et al, 1997; Short et al, 1999; Joiner et al, 2001) and by Williams and his colleagues (unpublished data). In some of these studies, the rate of cell kill can be more effective for dose-rates as low as 2 cGy/hr than for acute irradiation but only in some cell lines. In other cell lines this dose-rate is completely ineffective. Joiner and his colleagues suggest this response to low dose-rate correlates with cell susceptibility to induced hypersensitivity by low acute doses (< .05 Gy). Thus, there seems to us to be a strong imperative for further study of dose-rate as a possible important clinical tool, defining the molecular and cellular characteristics that will predict the response of tumor cells to lower dose-rates.

Selecting an appropriate dose rate and dose per fraction would, of course, be important. It has been proposed that a minimum dose-rate effect is most likely to occur within the range of 0.06 to 0.6 Gy/hr (Vilencheck and Knudson, 2000). Table 2 presents examples of studies in which an inverse dose-rate effect has been observed with protracted exposure to low-linear energy transfer (LET) radiations such as x-rays and g -rays.

Studies of low-dose total-body irradiation (TBI) with external beam sources suggest that additional benefit may be gained through up-regulation of immunological mechanisms that contribute to the overall anti-tumor effect. Pre-clinical reports indicate that low doses of TBI can significantly delay tumor growth (Anderson et al, 1982), decrease incidence of lung metastases (Hosoi and Sakamoto 1993), decrease incidence of spontaneous lymphoma (Ishii et al, 1996b) and increase the number of cells needed to produce a progressively growing tumor (Sakamoto et al, 1987). The low-dose radiation-initiated immune augmentation appears to be related to enhanced secretion of cytokines needed for immune cell activation (Sherman et al, 1991; Ishii et al, 1996a; Hosoi et al, 2001), increased expression of adhesion molecules that facilitate leukocyte trafficking (Hallahan, 1996; Hallahan et al, 1996, 1997) and preferential destruction of cells with immunosuppressive properties (Tilkin et al, 1981; North, 1986). The rationale for the latter mechanism is based partly on observations that infiltrating tumor-specific T cells are converted from a radiosensitive to a more radioresistant state due to activation by tumor antigens (Dunn and North 1991). Activated lymphocytes have long been known to be more resistant to radiation than their resting (non-activated) counterparts (Anderson and Warner, 1976). Further investigation in this area may contribute to greater efficacy of both localized and more general application of low-dose LDR radiotherapy. It has even been proposed that a radiation regimen could be implemented to enhance response to tumor antigens in individuals immunized with cancer vaccines (Friedman, 2002).

Interstitial and intracavitary implantation of radioactive sources (i.e. brachytherapy) was first practiced in the early 1900s and, with the exception of gynecologic malignancies, abandoned owing to poor results. However, recent improvements in placement and imaging have generated renewed interest in this form of radiotherapy. Brachytherapy is an extremely important model to consider the effectiveness of LDR irradiation. Brachytherapy delivers a mixture of dose-rates from HDR to LDR to VLDR. It can deliver very high total doses that are supra-lethal to tumor volumes immediately contiguous to implanted sources and failure of brachytherapy is usually attributed to tumor volumes on the periphery of the irradiated volume that escape lethal exposure. Although peripheral volumes receive essentially lower dose-rate irradiations (Lichter, 2000), there is little or no evidence to indicate that susceptibility of the peripheral cells to acute external-beam radiation treatment is altered.

Radioimmunotherapy, involving infusion of radiolabelled monoclonal antibodies (Mab), is experiencing renewed interest. Many Mab conjugated to radionuclides such as ¹¹¹In and ^99mTc have already been approved by the Food and Drug Administration (FDA) for tumor imaging and conjugation to high-energy therapeutic radionuclides such as ¹³¹I or ⁹⁰Y has resulted in some success. Thus, it appears that radiolabelled Mab could be exploited to concentrate low doses of LDR radiation to tumors (Knox et al, 1990, 1993; Kroger et al, 2001; Eriksson et al, 2003; Hernandez and Knox 2003) and to induce low dose-rate effects that would increase the efficacy of focal external beam therapy. It is clear that a better understanding of the key mechanisms responsible for both tumor and normal cell responses to low-dose protracted irradiation is needed before widespread clinical application can be seriously contemplated. This most

Table 2. Studies in which an inverse dose-rate effect has been reported after low-LET irradiation

certainly also holds true for the immune-related mechanisms that are triggered by radiation exposure.

As noted above, evaluation of dose-rate effects on the survival of tumor cells began more than 40 years ago. Some of the early findings are described in detail in publications from the 1980s and 1990s (Kelland and Steel, 1986; Steel et al, 1987; Steel, 1991, 1996). Dose rate dependence has been noted for some, but not all, of the malignant cell populations that have been tested. The reported variability and sometimes contradictory results make generalizations difficult. Selected relevant publications are summarized on the basis of tumor type in the following sections.

In 2004, 18,400 new cases of brain and other nervous system cancers and 12,690 deaths were estimated to have occurred in the United States (Jemal et al, 2004). Although not among the most common neoplasms, high-grade malignant tumors of the brain are among the most deadly and also among the most radioresistant type of cancer. The radioresistance has been well documented in reports of local recurrence within the targeted volume even after relatively high radiation doses (Wallner et al, 1989). Proposed mechanisms that may account for the resistance include radiation-induced up-regulation of antioxidant enzymes within the tumor cells (Lee et al, 2004), overexpression of epidermal growth factor receptors (EGFR) (Barker et al, 2001) and low expression of bax, a pro-apoptotic protein (Shu et al, 1998; Streffer et al, 2002). In addition, the possibility of tumor escape from immune surveillance by the production of transforming growth factor-b 2 (TGF-b 2), a highly immunosuppressive cytokine that is secreted by the majority of glioblastomas, is under investigation (Kingsley-Kallesen et al, 2001; Strege et al, 2004). Although TGF-b 2 was not tested for, the pattern of cytokine mRNA expression in five human glioblastoma cell lines has been shown to be dependent upon radiation dose rate (Ross et al, 1997). Dose rates of 0.0035 Gy/min and 0.01 Gy/min generally reduced the mRNA levels for IL-1b and IL-6 compared to unirradiated control cells. On the other hand, exposures at 0.041 Gy/min and 2 Gy/min increased mRNA for both cytokines. This study also showed that the 0.041 Gy/min intermediate dose rate was less toxic than the lower and higher dose rates. Although cytokine gene expression did not have a clear effect on glioblastoma cell survival in this in vitro study, the investigators speculated that tumor-derived cytokines may be important to radiation response in vivo by affecting immune cells, tumor stroma, vasculature, or surrounding tissues.

In a follow-up publication, the effect of ultrafractionated radiation on A7 glioma growing subcutaneously (s.c.) in athymic nude mice was determined (Krause et al, 2003). The A7 cells were among those that had previously shown a striking increase in radiosensitivity with ultrafractionated irradiation in vitro. Tumors were irradiated either with 126 fractions at 0.4 Gy/fraction or with 30 fractions at 1.68 Gy/fraction over a period of 6 weeks. A total dose of 50.4 Gy was delivered to both groups of animals. No increase in radiosensitivity was evident with ultrafractionation. In fact, a significant decrease in tumor growth delay occurred with ultrafractionation compared to the more conventional 1.68 Gy/fraction. The investigators noted that simplistic extrapolation from data obtained in vitro is not sufficient to predict the outcome in vivo and that comprehensive evaluation of new treatment options in animal models is essential before contemplating translation to the clinic.

Other investigators have also noted lack of correlation between human glioma cell lines when exposed to fractionated radiation schemes in vitro and when treated as xenografted tumors in athymic rodents (Baumann et al, 1992), as well as increasing glioma cell survival as the radiation dose rate is reduced (Yang et al, 1990). In one of these studies, Yang and colleagues evaluated 16 clones of an early-passage human glioma cell line (IN859) based on DNA content, modal chromosome number, morphology and radiosensitivity (Yang et al, 1992). The radiation dose that gave a survival fraction of 0.01 varied by a factor of ~1.5. Comparison of the most sensitive and most resistant clones at a fixed total dose surprisingly demonstrated that the sensitive clone exhibited greater survival after split-dose irradiation. Finally, with T98G glioblastoma cells, a sparing effect was noted when the dose rate was reduced from 60 cGy/hr to 30 cGy/hr (Mitchell et al, 2002).

The above findings, although somewhat disappointing, certainly do not disprove the existence of low-dose hyper-radiosensitivity for tumors originating in the CNS. In a very recent study by Chalmers and colleagues, the effects of four different poly(ADP-ribose) polymerase-1 (PARP-1) inhibitors on low-dose (0.5 — 0.3 Gy) radiosensitivity were evaluated using T98G and U373-MG human glioma cell lines (Chalmers et al, 2004). Hamster fibroblasts (V79-379A and CHO-K1) and mouse embryo fibroblasts (3T3) were also included in this study. Inhibition of PARP-1 resulted in sensitization of the glioma and most of the other cell lines to the low-dose radiation; the only exception was the mouse 3T3 fibroblast line (i.e. PARP-1 knockout cells). The investigators suggested that PARP-1 inhibitors may be therapeutically valuable in radiotherapy regimens that consist of multiple small doses or continuous LDR radiation because of the enhanced inhibition of PARP-1 on rapidly dividing tumor cells. PARP-1 is an enzyme that binds rapidly to single- and double-stand DNA breaks and modulates the activity of many nuclear proteins (D'Amours et al, 1999; Herceg and Wang, 2001). Deficiency in PARP-1 results in severely impaired base excision repair and genomic instability in response to low-dose radiation (Shall and de Murcia, 2000).

An encouraging in vivo study by Williams and colleagues combined ULDR radiation together with HDR radiation using athymic mice that were s.c. implanted with U251 human malignant glioma cells (Williams et al, 1998). The ULDR radiation was delivered locally as intratumorally implanted ¹²⁵I seeds (0.05 Gy/hr) or from an external ¹³⁷Cs source (0.037 Gy/hr) that provided whole-body exposure. The external beam HDR radiation treatments were initiated 3 days later and consisted of 2 Gy x 8 daily fractions and 5 Gy x 2 daily fractions. These time-dose regimens were selected because they are clinically feasible and because previous experiments predicted equal growth delay of the tumor. The ULDR radiation alone had little or no effect on tumor progression and the two HDR external beam schedules resulted in a 20-25 day growth delay. However, the continuous administration of ULDR radiation with either ¹²⁵I seeds or whole-body exposure to ¹³⁷Cs significantly increased efficacy of the HDR treatments. A delay of 33-35 days in growth of the glioma was obtained with the combination treatments.

Adenocarcinoma of the prostate is the leading cancer diagnosed and the second most common cause of cancer-related mortality in males residing in the United States. According to estimates of the American Cancer Society, there were approximately 230,110 newly diagnosed cases of prostate cancer and 29,900 deaths due to the disease in 2004 (Jemal et al, 2004). Most patients presenting with prostate cancer have localized disease and are candidates for radiation therapy. Exposure to fractionated HDR radiation is a standard treatment for localized prostate cancer. With conventional radiation time-dose regimens, historical local recurrence rates range from 25% to 62%, leaving considerable room for improvement. Protocols that increase local control are likely to be very important in mitigating the subsequent development of distant metastases and prolonging survival (Fuks et al, 1991; Kaplan et al, 1994). The still relatively high incidence of side effects (e.g. proctitis, rectal bleeding, increased urinary frequency, urethral stricture and gastrointestinal complications) is also a major concern.

Studies describing tumor cell responses to LDR/HDR radiation combinations present the possibility that radiotherapy of prostate cancer could be substantially improved. Most prostate cancer cell lines require relatively large doses of HDR radiation to produce significant cell death (Wollin et al, 1989; Smalley et al, 1991; Kaver et al, 1991; Leith et al, 1993; Leith, 1994). Prostate cancer cells subjected to a single acute dose of 2 Gy (a commonly used dose/fraction in prostate cancer radiotherapy) generally have a survival fraction of approximately 0.56 (DeWeese et al, 1998), whereas the survival fraction of the more radiosensitive lymphomas is often <0.35 under similar conditions (Jones et al, 1973; Malaise et al, 1986).

In the United States, cervical carcinoma and other cancers of the female genital system accounted for about 82,550 new cases and 28,720 deaths in 2004 (Jemal et al, 2004). High-grade, advanced and/or recurrent gynecological malignancies tend to have a poor prognosis. They are difficult to cure with conventional external irradiation or any other therapeutic modality. Thus, patients with gynecological cancers are often selected for HDR, pulsed-dose-rate and other variations of brachytherapy (Jensen et al, 1998; Lessard et al, 2002; Rose, 2003). A regimen that includes low-dose protracted exposure has reportedly been beneficial in at least some of these cases. Increasing knowledge from pre-clinical studies, some of which are described below, should eventually help refine these protocols in patients.

Asynchronous and synchronized cervical carcinoma cells have been exposed to ⁶⁰Co g -rays at 0.33 and 0.86 Gy/hr (Furre et al, 1999). Clonogenic survival data indicated the same degree of radiation sensitivity for exposure periods of less than 20 hr. However, with exposures of >20 hr, a 2-fold greater radiosensitivity was noted and was found to occur when 80% of the cells had accumulated in G2. Thus, in this study of cervical carcinoma cells, an inverse dose-rate effect was demonstrated (i.e. more efficient cell inactivation at lower compared to higher dose rates) when total radiation doses exceeded 7 Gy. Other investigators have also demonstrated that the total dose delivered over an extended time period can influence radiosensitivity (Lamerton and Lord, 1964; Hall et al, 1966).

Dose-rate effects on human adenocarcinomas of the uterine cervix (NHIK-3025, HeLa, HeLa S3) and human squamous cell carcinomas (Me180 from the cervix and A431 from the vulva) have been investigated in xenotransplanted athymic mice (van Oostrum et al, 1990). Intratumorally implanted cesium-157 (¹⁵⁷Cs; 117 MBq) needles 15 mm in length were used to deliver radiation at 0.5 Gy/hr. Tumor biopsies were taken from 21-160 hr after radiation initiation. One biopsy was taken from tissue that received a 10 Gy dose, whereas other biopsies were taken toward the tumor periphery that had received total doses of 2-9 Gy. Tumors were also exposed to a much higher dose rate of 3-4 Gy/min to total doses ranging from 3-10 Gy using a linear accelerator. Biopsies of these tumors were taken at times corresponding to biopsy times after the 0.5 Gy/hr irradiation. Changes in tumor cell cycle distribution were determined by flow cytometry analysis. All five tumor cell lines were arrested in G2 after the 3-4 Gy/min exposure, whereas only four were arrested in G2 after 0.5 Gy/hr irradiation. Me180 was the exception that accumulated primarily in the S phase. Maximum G2 accumulation was related to cell cycle time and not the dose. Both irradiation conditions resulted in similar changes in cell cycle progression, but a greater maximum G2 accumulation was observed with the higher dose-rate exposure. In a separate experiment, the investigators showed that the enhancement of cells in G2 correlated with radiosensitivity as determined by measurements of delay in tumor regrowth.

In another study, the radiation response of three cervical carcinoma cell lines (HX155c, HX156c and HX160c) was compared in vitro and in vivo as xenografted tumors (Tonkin et al, 1989). ⁶⁰Co g -rays were delivered at continuous rates of 0.03 — 0.05 Gy/min and 0.7 — 1.0 Gy/min. Two of the three cell lines showed significant low-dose sparing in vitro and there was a tendency for the in vivo tumors to reflect a similar pattern. However, there was less tumor growth delay in vivo than that predicted from the in vitro data.

The effects of protracted radiation delivered by Mabs conjugated to radionuclides are being explored in combination with HDR external beam radiation in pre-clinical studies of gynecological tumors. This technique is also known as radioimmunotherapy. Success with this combination treatment was demonstrated in a recent study by Eriksson and colleagues utilizing athymic mice with s.c. implanted human HeLa Hep2 cervical carcinoma cells (Eriksson et al, 2003). Treatments with external beam radiation at 3 x 5 Gy and/or 100 m g ¹³¹I-labelled Mab against placental alkaline phosphatase or ¹³¹I-Mab against cytokeratin were administered separately or in combination; specific activity of the Mab was 120-200 Mbq/mg antibody. Although tumor growth retardation was observed with both external beam and the radiolabelled Mab, combination treatment enhanced the therapeutic effects further, resulting in a significant reduction in tumor volume. The long-lasting tumor growth inhibition was related to increases in tumor necrosis and apoptosis. The investigators suggest that this combination approach could increase therapeutic efficiency for epithelial cell-derived tumors in general.

Approximately 110,960 individuals were newly diagnosed and 55,100 deaths occurred in the United States in 2004 owing to hematological malignancies (Jemal et al, 2004). Collectively, leukemia, lymphoma and multiple myeloma represent the fourth most common form of cancer. Current radiation and multiple-agent chemotherapy regimens can be curative in a substantial percentage of patients with certain forms of these diseases. Nonetheless, more children die of leukemia than any other disease and the death rates for non-Hodgkin's lymphoma and multiple myeloma are increasing in the U.S.A. It seems possible that implementation of LDR radiation may be beneficial in at least some of these cases.

In some studies of hematological malignancies, the apoptotic effect of protracted radiation has been demonstrated to be independent of p53 status. In one of these investigations, the Raji model that mimics therapy-resistant human lymphomas with mutant p53 and increased bcl-2 expression was used (Kroger et al, 2001). Athymic nude mice bearing Raji lymphoma xenografts were treated with ⁶⁷Cu-2IT-BAT-Lym-1 antibody (335-500 m Ci) and observed for toxicity and tumor response for 84 days. Subsets of animals were euthanized at 3, 6 and 24 hr after therapy so that tumors could be examined for evidence of apoptosis and for p53, bcl-2, p21, GADD45, TGF-b 1 and c-myc gene expression and protein level. Apoptosis was greatly increased in the treated xenografts, whereas bcl-2 gene and protein expression were substantially decreased. These changes occurred despite only modest cumulated radiation doses of approximately 0.56 Gy at 3 hr. Apoptosis preceded tumor regression by 4-6 days and 29% of the tumors were cured by cumulated tumor doses of ~18 Gy.

Amundson and colleagues have studied gene expression patterns following LDR and acute g -irradiation of the human ML-1 myeloid leukemia cell line (Amundson et al, 2003). ML-1 cells have wild type p53 and are relatively responsive to ionizing radiation in that gene expression changes have been induced with as little as 0.1 Gy. The study evaluated a wide range of genes by cDNA microarray after irradiating the ML-1 cells at dose rates of 0.0028 Gy/min to 2.9 Gy/min, with total doses ranging from 0.02 Gy to 0.5 Gy. Two major clusters of genes were radiation-induced. The majority of genes in the dose rate-independent group are known to be involved in cell cycle regulation, whereas the majority of those in the dose rate-dependent cluster have roles in apoptosis. The findings were consistent with the significant decline in the percentage of apoptotic cells with decreasing dose rate that was observed with fluorescence microscopy. In spite of this, the investigators stated that for the majority of genes responding to low doses of radiation, a protective dose-rate effect does seem to apply in the case of this human myeloid leukemia cell line. A slight, but statistically significant, increase in survival has also been reported for LX830 mouse leukemia cells with dose rates decreasing from 30 mGy/hr to 6.2 mGy/hr (Furuno-Fukushi and Matsudaira, 1989). In this latter study, no differences in mutation frequency were associated with dose rate. Other investigators have found little or no dose-rate effects on survival of repair-deficient mutants such as mouse lymphoma LYS-s, xrs5, xrs6 and irs20 cells (Evans et al, 1985; Nagasawa et al, 1989; Stackhouse and Bedford, 1993).

In a study by Williams and co-workers, four human tumor cell lines were evaluated in colonial survival assays after exposure to protracted irradiation: colon (GEO and LS174T), squamous cell (SQ-20B) and hepatocellular (HepG2) carcinoma (Williams et al, 1992). Analysis of cell survival data was performed using a 7-parameter simulation model (Dillehay, 1990). Tritiated water added to cell culture media was used for ULDR irradiation according to the formula 0.08 mCi/ml = 0.01 Gy/hr, whereas HDR radiation at 1.1 Gy/min was administered using a g -ray irradiator. Data for three of the cell lines (GEO, LS174T and SQ-20B) showed diverging rates of cell kill per unit dose of HDR radiation. With protracted irradiation, inactivation rates diverged at low doses, but tended to converge at higher doses. The HepG2 cells, tested only with LDR radiation at 0.03 Gy/hr, exhibited no redistribution in the cell cycle after 24, 48, or 72 hours of exposure.

Neuroblastoma is a common malignancy of early childhood, most often arising from the adrenal medulla. A study of two human neuroblastoma cell lines (HX138 and HX12) demonstrated no dose-rate effects above 0.02 Gy/min (Holmes et al, 1990). In addition, when the dose rate was reduced below 0.02 Gy/min, both cell lines exhibited increased survival.

Clonogenic survival and double-strand DNA breaks were examined in the RT112 human bladder carcinoma cell line using radiation dose rates ranging from 0.01 Gy/min to 1.28 Gy/min (Ruiz de Almodovar et al, 1994). Immediately after irradiation, cell survival increased with decreasing dose rate, as did the number of double-strand breaks measured by pulsed-field gel electrophoresis. However, when a 4-hour repair period was allowed after exposure, all dose rates resulted in approximately the same amount of damage. The investigators concluded that the level of un-rejoined double-strand DNA breaks did not correlate with survival at different dose rates, when detected by this particular gel electrophoresis technique.

Tumor cells vary in terms of their radiosensitivity dependence on cell cycle distribution, as well as expression of pro-apoptotic genes (Grdina, 1980; Keng et al, 1984; Brock et al, 1987). In the case of some tumor cell types, the degree of radiosensitivity has been associated with levels of pro-apoptotic proteins such as p53 and bax. Ohnishi and collaborators found that cultured human squamous carcinoma cells primed with g -radiation at 0.001 Gy/min to a total dose of 1.5 Gy and then challenged with g -rays at 1 Gy/min exhibited decreased apoptosis and depressed accumulation of p53 and bax (Ohnishi et al, 2001).

Induction of micronuclei was used as an indicator of apoptosis in Lewis lung carcinoma (LLC) subjected to g -radiation dose rates of 0.34 Gy/min and 1 Gy/min while growing in mice (Widel and Przybyszewski, 1998). Tumors were excised, cultured in vitro and scored for micronuclei at 24-hour intervals. In the range of 0 to 6 Gy, the frequency of cells containing micronuclei was linearly dependent on dose. However, an inverse dose-rate effect was evident, with the lower dose rate resulting in a higher frequency of micronuclei per cell than the higher dose rate. Since the difference in exposure times between the two dose rates was not great, the investigators speculated that a differential radiation effect on cell kinetics during tumor irradiation could not explain the results. Rather, it appeared possible that the differential effect was dependent upon division delay and redistribution of cells in the phases of the cell cycle during in vitro incubation. In other words, they hypothesized that more cells in the higher dose-rate group died in culture because of interphase death than in the lower dose-rate group and hence the cells were no longer available for the micronucleus assay.

Normal cells respond to ionizing radiation with a delay in progression through the cell cycle so that DNA damage can be repaired before DNA replication and mitosis take place. The delay decreases the chance that genomic instability or mutant phenotypes will appear among the cell progeny. However, if the damage is too extensive, the cell will be eliminated because of post-mitotic or apoptotic death. Surviving cells reenter the cell cycle after various lengths of time, depending upon the radiation dose and the specific cell type involved. The delay occurs primarily in the G2 phase, although radiation can also delay cells in G1 and S. Expression of the TP53 tumor suppressor gene is especially important in these events (Grdina, 1980; Keng et al, 1984; Rauth, 1992; Bristow et al, 1996). If TP53 is lacking or is functionally deficient, survival of cells with DNA damage and thus also risk for subsequent malignancy, is increased (Kato et al, 2002). However, the degree of radiosensitivity varies substantially among normal cell populations, especially under conditions of protracted low-dose exposure.

With conventional radiotherapy protocols, both acute and late toxicities are possible (Lichter, 2000; Yamada et al, 2000; Gopal et al, 2001). Acute effects, seen during and up to 3 months following the end of treatment, occur primarily in rapidly proliferating tissues such as the skin, bone marrow and mucosa of the oropharynx, gastrointestinal tract, rectum and bladder. Erythema, esophagitis, pneumonitis, diarrhea, dysuria and leukopenia are among the most common manifestations. Late effects occur 3 months or longer after the end of radiotherapy and can involve virtually any organ or tissue. Most late reactions include progressively increasing fibrosis, blockage of blood vessels resulting in tissue anoxia and/or ulceration. Late effects are generally more serious and more difficult to manage than the acute toxicities. Eventually, death may occur due to organ failure or other complications. Development of secondary tumors within the irradiated volume is another consequence of radiotherapy that is likely to become more prevalent with increasing survival of cancer patients (Schneider et al, 2000; Strojan et al, 2000; Kranzinger et al, 2001).

Injury to normal lung tissue and the gastrointestinal tract (GI) remains a major problem after total—body irradiation (TBI) in bone marrow recipients with refractory hematological malignancies and serious immuno-deficiencies (e.g. severe combined immunodeficiency disease). It was proposed several decades ago that radiation-induced lung toxicity may be ameliorated by the use of continuous protracted or fractionated radiation, since repair of sublethal damage is greater for the lung and GI tract than for hematopoietic tissue and probably also leukemic cells (Dutreix et al, 1981).

Many preclinical studies have employed localized lung irradiation, rather than TBI. In a study with mice, radiation was delivered to the upper body at dose rates ranging from 0.02 to 1.0 Gy/min (Down et al, 1986). A continuous increase in normal tissue tolerance was observed for early radiation-induced pneumonitis with decreasing dose rate, but a less pronounced sparing effect was noted for late complications. To evaluate the effects of adriamycin on radiation-induced pulmonary and upper GI tract toxicity, the thoracic region of mice was irradiated at dose rates of 0.05, 0.15 and 0.70 Gy/min (Sherman et al, 1982). The beneficial effect of the lower dose rates was markedly diminished, although still evident when the drug was administered pre-irradiation. In other words, this study showed that adriamycin can significantly increase the oral esophageal and pulmonary toxicity of radiation and can almost abrogate the sparing effect of dose rate. In a mouse model of bone marrow transplantation, Safwat and colleagues compared TBI delivered at a rate of 0.08Gy/min with TBI administered at a rate of 0.71 Gy/min, both with and without cyclophosphamide (Safwat et al, 1996). The drug was administered 24 hr before irradiation and the transplant was performed 4-6 hr after the last treatment. Lung damage, assessed using ventilation rate and mouse lethality, was decreased using the lower dose-rate regimen. However, administration of the drug markedly facilitated the TBI-induced damage. Somewhat unexpectedly, the combination of protracted radiation plus drug was found to be more toxic than acute radiation plus drug. In another study, mice given cyclophosphamide before thoracic irradiation at dose rates ranging from 0.05 Gy/min to 1.0 Gy/min developed pneumonitis at 4-9 weeks post-exposure (Lockhart et al, 1986). Without the drug, pneumonitis appeared at 14-16 weeks, regardless of radiation dose rate. In a rat model of TBI, fractionation had little effect on bone marrow ablation, but resulted in increased gastrointestinal and renal tolerance (Moulder and Fish, 1989).

Results from studies in patients receiving various TBI regimens in association with bone marrow or stem cell transplantation have generally supported use of LDR or increased fractionation as a means to minimize normal tissue toxicities, increase the total dose of radiation that can be safely delivered and improve tumor control (Evans, 1983; Regnier, 1992; Corvo et al, 1999; Gopal et al, 2001; Song et al, 2003). Radiation therapy at low doses is now commonly employed in cases of chronic lymphocytic leukemia and low-grade non-Hodgkin's lymphoma. The radiation is administered to the entire body at a very low dose per fraction (0.1 to 0.25 Gy) several times a week until a total dose of approximately 1.5 to 2 Gy is reached. Interestingly, the anti-tumor effects have been reported to be more pronounced than what would be expected from direct tumor cell destruction by radiation. It appears that immune enhancement is among the mechanisms by which TBI results in long-term remissions in the majority of patients. Pre-clinical studies have demonstrated that TBI up-regulates immune responsiveness in a number of ways: a) increases IFN-g and IL-2 production; b) increases expression of IL-2 receptors on T lymphocytes; c) enhances signal transduction in T cells; d) augments T cell proliferation in response to mitogens; e) lowers serum corticosterone and increases catecholamine in spleen; and f) eliminates T cells with immunosuppressive activity (Anderson et al, 1982, 1988; Safwat, 2000a, 2000b).

Low-dose TBI has been recently tested in combination with IL-2. In one study, mice were inoculated i.v. with B167F1 malignant melanoma cells on day 0 (Safwat et al, 2003a). A single whole-body dose of 0.75 Gy was administered on day 7; IL-2 was initiated on day 8 and given twice daily for 5 consecutive days. Pulmonary tumors were quantified and examined for tumor-infiltrating cells on day 14. In groups treated with either modality alone, tumor burden was similar to that in non-treated controls. However, the combination of low-dose TBI and IL-2 resulted in a synergistic anti-tumor effect. Natural killer (NK) cells and macrophages were identified as the most likely participants in the highly significant outcome of combination treatment. In a subsequent study of malignant melanoma, these same investigators used two doses of TBI and IL-2 (Safwat et al, 2004). Combining TBI with high-dose IL-2 led to a further significant reduction in tumor and less severe vascular leakage syndrome compared to high-dose IL-2 alone. In this case, therapeutic efficacy was associated with the number of tumor-infiltrating NK cells. The investigators concluded that the combination treatment was not only more effective, but also less toxic than IL-2 alone. Although very few human studies have been conducted in this area, data suggest that low-dose radiation-induced immune defense mechanisms may be operating (Safwat 2003b). Based on these types of data, some have suggested that administration of low-dose TBI may be beneficial for patients with AIDS (Shen et al, 1989; Shen et al, 1997). In contrast, TBI has also been associated with enhanced tumor progression (Duhrsen and Metcalf, 1988; Gridley et al, 1997). The discrepancies in reported data may be related to tumor type and load, as well as the dose rates and total doses of radiation utilized.

Radiotherapy for neoplasms of the brain inevitably results in some damage to neurons. The damage can trigger a series of events, leading to destruction and malfunction of a relatively large volume of tissue due to secondary degeneration (Ikonomidou and Turski, 1996; Yoles and Schwartz, 1998). Side effects ranging from headaches, dizziness, nausea and cognitive defects to development of secondary CNS neoplasms following treatment remain problematic and are likely to become more obvious with increased survival times (Leibel SA and Sheline 1987; Schultheiss et al, 1995; Strojan et al, 2000; Kranzinger et al, 2001). The success of anti-inflammatory drugs such as corticosteroids in controlling edema and increased intracranial pressure suggests that cells of the immune system play a role in the development of at least some of the observed toxicities. Corticosteroids have a number of non-specific anti-inflammatory and immunosuppressive properties including: a) they inhibit neutrophil migration to sites of tissue damage and thus minimize the level of neutrophil-derived vasoactive amines that increase vascular permeability and produce edema; b) they suppress IL-1 production by cells of the monocyte/macrophage lineage, a cytokine that is needed for T cell activation: and c) they impair synthesis and secretion of IL-2 by T lymphocytes, which in turn reduces the ability of the T cells to proliferate and produce pro-inflammatory cytokines.

For many decades, the CNS was considered to be an immunologically privileged site, with little or no interaction occurring between the immune and central nervous systems; if immune activity did occur, it was thought to be detrimental (Streilein, 1993; Popovich et al, 1996). However, a series of recent studies have demonstrated that recruitment of systemic T lymphocytes to the injured site helps the innate branch of the immune system to ward off toxicity (Moalem et al, 1999; Schwartz and Cohen, 2000; Fisher et al, 2001; Kipnis et al, 2001, 2002). In contrast to healthy tissue, the injured CNS is more accessible to circulating lymphocytes because of blood-brain barrier breakdown and expression of adhesion molecules, chemokines and major histocompatibility complex (MHC) class II molecules. The in-migrating T cells belong to the Th1 subset that secretes cytokines such as IL-2 and IFN-g . In addition, they are autoreactive and respond to specific 'self' antigens (e.g. anti-myelin protein) that are presented to them by activated microglia. Indeed, it has been demonstrated that rats and mice that are either deficient in mature T cells or that lack auto-reactive T cells lose their ability to withstand injury to the CNS (Kipnis et al, 2000; Schori et al, 2001). The existence of a protective role for the immune system has been strengthened by the demonstration that passive transfer of a subpopulation of T suppressor cells (i.e. the naturally occurring regulatory CD4⁺/CD25⁺ T cells) can eradicate irradiation-induced Th1 cell-derived neuroprotection (Kipnis et al, 2002). Most recently, a study was done to determine whether g -radiation benefits neuronal survival by alterating immune system status (Kipnis et al, 2004). Mice and rats were subjected to optic nerve crush or contusive injury to the spinal cord. Total-body or total-lymphoidal irradiation (3.5 Gy) resulted in significant increases in neuronal survival (in some cases more than 3-fold) and more rapid recovery from injury. The beneficial effects were noted when radiation was administered up to 3 days post-injury.

Numerous studies of dose-rate effects on mutagenesis have been performed using human and rodent lymphoid cells (Lorenz et al, 1993; Lorenz et al, 1994; Amundson and Chen, 1996; Furuno-Fukushi et al, 1996). The hypoxanthine-guanine phosphoribosyl transferase (HPRT) and thymidine kinase (tk) loci have arguably received the greatest attention. In a study of mutation size at the HPRT locus in human lymphoblastoid WI-L2-NS cells, an inverse dose-rate effect was found after a total dose of 4 Gy (Colussi and Lohman, 1997). Significantly larger deletions were produced by LDR exposure than with HDR exposure. Since the majority (>60%) of cells irradiated with HDR were in G1 and those irradiated with LDR went through the S-phase and on into G2, the investigators proposed that the difference in mutation size may be related to the stage of the cell cycle. Other studies have demonstrated that decreasing the dose rate results in an inverse dose-rate effect as manifested by increased mutagenesis (Furuno-Fukushi et al, 1988; Crompton et al, 1990; Amundson and Chen, 1996). In contrast, some investigators have reported that protracted radiation exposures protect against mutation induction both in vitro (Evans et al, 1990) and in vivo (Lorenz et al, 1994) or have no significant effect compared to acute irradiation (Furuno-Fukushi et al, 1996). In some cases, discrepancies have been observed between in vitro and in vivo data (Lorenz et al, 1993). Reports by Furuno-Fukushi and colleagues using non-transformed, near-diploid m5S mouse cells demonstrated that cell proliferation during protracted irradiation has a strong influence on mutagenesis (Furuno-Fukushi et al, 1993). In plateau-phase cultures, lowering the dose rate from 30 Gy/hr to 13 mGy/hr resulted in an increase in cell survival and a marked decrease in mutation frequency. However, when the cells were in log-phase culture, the magnitude of the dose-rate effect was not nearly as marked, especially when the total dose was below 5 Gy. An interesting study of mutagenesis was recently performed in which chronic low-dose rate g -irradiation was used to mimic the environment experienced by residents during the Chernobyl accident (Wickliffe et al, 2003). Big Blueâ mice were exposed to g -rays for 90 days, resulting in a cumulative dose of 3 Gy (0.0014 Gy/hr). No significant increase in mutation frequency was observed in the irradiated mice based on DNA analysis of liver tissue (Big Blueâ Transgenic Rodent Mutagenesis Assay System).

The above-mentioned studies, as well as others, indicate that variations in experimental conditions may at least partly account for the apparent contradictions that have been noted in radiation-induced mutagenesis. Overall, the preponderance of evidence indicates that the risk for radiation-induced mutagenesis in somatic and germ line cells decreases with decreasing dose rate (Vilencheck and Knudsen, 2000).

The consensus that radiation is harmful to normal cells has prevailed for many decades. Among the most serious consequences are immunodeficiency, mutations and development of malignancies such as leukemia and multiple myeloma (reviewed by Dainiak 2002). Numerous studies of radiation-induced chromosomal and DNA aberrations have been performed on lymphocytes and lymphoid cells (Hofer et al, 1994; Kronenberg, 1994; Moiseenko et al, 1997; Wu et al, 1997, 1999; Blakely and Kronenberg, 1998; Durante et al, 1998, 1999, 2000, 2002; Yamada et al, 2000; Gauny et al, 2001; Grosovsky et al, 2001; Holl et al, 2001; Wiese et al, 2001; Schulz-Ertner et al, 2002). These studies have revealed important radiation-induced genomic abnormalities, as well as variations due to differences in radiation quality. Although lymphoid organs have long been known to be highly radiosensitive, the susceptibility of different lymphocyte populations varies greatly, possibly owing to differences in repair mechanisms, maturation stage, activation state and/or innate rate of proliferation (Stefani and Schrek 1964; Miller and Cole, 1967; Anderson and Warner 1976; Anderson et al, 1977, 1988; Prasad 1995). An additional important observation is that irradiation can up-regulate many genes, including some that encode cytokines that can modify cellular responses to radiation (Hallahan et al, 1993; Hallahan, 1996; Barcellos-Hoff, 1998;).

It is clear that radiation, when delivered at a slow continuous rate or by fractionation may have strikingly different effects compared to the same dose delivered acutely. Early studies demonstrated that monkeys could tolerate whole-body exposures up to 19 Gy with minimal suppression of the hematopoietic system when the radiation was delivered at a rate of 1 Gy/year (Spalding et al, 1972). The 19 Gy is approximately three times the lethal dose when applied acutely. Soon thereafter, it was reported that protracted radiation exposure may induce beneficial effects (i.e. radio-adaptation, previously known as 'hormesis') (Anderson and Warner, 1976; Anderson et al, 1977). The radio-adaptive response has gained much attention in recent years because of its obvious importance in setting accurate guidelines for radiation-associated human health risks, in basic radiobiology research and now also in radiation therapy. The phenomenon has been extensively discussed in a number of publications (Bhattacharjee and Ito, 2001; Feinendegen 2003; Pollycove and Feinendegen, 2003). In this review, only a few highlights of special interest are mentioned.

Various endpoints have been used to detect and quantify radio-adaptation, including chromosome aberrations (Shadley et al, 1987; Wolff et al, 1988; Sankaranarayanan et al, 1989; Farooqi and Resavan, 1993), mutations and DNA strand breaks (Shadley and Wolff, 1987; Kelsey et al, 1991; Rigaud et al, 1993), micronucleus formation (Azzam et al, 1994) and cell survival (Olivieri et al, 1984; Yoshida et al, 1993; Sasaki, 1995). In many of these studies, low-dose irradiation of lymphocytes has been found to minimize the harmful effects of a subsequently delivered high dose of radiation. For example, although lymphocytes from individuals occupationally exposed to chronic doses of radiation have higher frequencies of spontaneous micronuclei than non-exposed individuals, after 1 and 2 Gy irradiation of the lymphocytes in vitro this frequency is lower for the radiation workers (Gourabi and Mozdarani, 1998). It has also been reported that low-dose irradiation can enhance the production of IL-1 by lipopolysaccharide (LPS)-stimulated splenocytes, responsiveness to T cell mitogens (Ishii and Watanabe 1996) and expression of IL-2 receptors on the surface of peripheral blood lymphocytes (Xu et al, 1996). Many of these types of reports indicate that a significant radio-adaptive response is measurable within a few hours after low-dose priming (Farooqi and Resavan, 1993; Ishii and Watanabe, 1996; Park et al, 2000; Venkat et al, 2001).

With intact mammalian models, radio-adaptation in immune cells has been shown to protect mice injected with the Friend leukemia virus, a member of the retrovirus family that includes HIV-1 (Shen et al, 1989; Wolff et al, 1989; Fujiki and Suganuma, 1994). It has also been noted that the onset of thymic lymphoma is significantly delayed in Swiss mice pre-conditioned with 0.01 Gy and then acutely irradiated with 2 Gy (Bhattacharjee, 1996). Similar results have been obtained for acute myeloid leukemia (Mitchell et al, 1999) and some transplanted solid tumors in mice (Bhattacharjee and Sarma, 1999). In one very recent report, mice were subjected to single low-level exposures of 0.1 Gy or 0.2 Gy X-rays and injected i.v. 2 hr later with syngeneic L1 sarcoma cells (Cheda et al, 2004). Pre-treatment with either dose of radiation significantly reduced the number of tumor colonies in the lungs compared to the non-irradiated counterparts. In addition, significantly enhanced NK cell cytotoxic activity was observed in spleens from the irradiated animals and elimination of the NK cells by injection of anti-asialo GM₁ antibody totally abrogated the radiation-induced tumor-inhibitory effect. Studies by other investigators support these findings. Suppression of pulmonary nodule development with single doses of X-rays (0.05 Gy to 0.15 Gy) 24 hr before i.v. injection of B16 melanoma and Lewis lung cancer cells has been reported in mice (Cai, 1999). Hashimoto and co-workers have shown that lung and lymph node metastases are decreased and that metastatic foci are infiltrated by lymphocytes in rats exposed to 0.2 Gy g -rays 14 days after s.c. injection of hepatoma cells (Hashimoto et al, 1999). In this latter study, in vitro irradiation of the tumor cells or localized irradiation of tumor in vivo at the same dose did not affect either primary tumor growth or degree of spontaneous metastases. Irradiation of mice with X-ray doses of <0.2 Gy has also been reported to suppress local tumor growth when malignant cells are injected after the radiation exposure (Anderson et al, 1982; Cai, 1999). Some investigators have suggested that the enhanced anti-tumor effect may be related to secretion of TNF-a and other cytokines by cells of the monocyte-macrophage series that become activated by debris due to radiation-induced damage (Hallahan et al, 1989; Sherman et al, 1991; Weichselbaum et al, 1991). TNF-a may add to the lethality of radiation because it has both direct and indirect anti-tumor effects.

Although radio-adaptation was not specifically addressed, our previous studies have demonstrated that s.c. growth of Lewis lung carcinoma in C57BL/6 mice is consistently slower when the animals are exposed to 3 Gy g -rays 2 hr prior to tumor cell implantation (i.e. representing a small neoplastic focus that appears after the irradiation event has taken place) and that the anti-tumor effect was related to immune enhancement during repair of normal tissue damage (Miller GM et al, 2002, 2003a, 2003b). The data also showed that a dramatic expansion of NK cells occurred in the blood and spleen of tumor-bearing animals 10 days after irradiation (28.0 x 10⁵ NK cells/spleen in irradiated mice compared to 8.9 x 10⁵ NK cells/spleen in non-irradiated controls). Enhanced production of IL-12 and IL-18 by spleen cells was consistent with augmentation of the NK cell response. Significant reductions in TGF-b 1 and vascular endothelial growth factor (VEGF), both of which are associated with immune suppression, were also noted. Enhanced NK cell activity after total-body irradiation with 0.075 to 0.5 Gy X- or g -rays has been reported by other investigators (Kojima et al, 2002, 2004; Liu et al, 1994). The important role of NK cells in nonspecific immune surveillance against aberrant cells and metastatic growth has been well established (Talmadge et al, 1980; Hanna, 1985; Wiltrout et al, 1985; Moretta et al, 1994; Reyburn et al, 1997; Barao and Ascensao, 1998). These findings are consistent with the radiation-induced "danger" signal originally proposed by Matzinger in 1994 and recently reviewed by Friedman (2002). Based on studies such as those above, it has been suggested that low-dose whole-body irradiation may be useful in treating some cancer patients (Pollycove and Feinendegen, 2000). However, it should be mentioned that the incidence of radiation-induced carcinogenesis increases in humans when doses exceed approximately 0.3 Gy (Pollycove and Feinendegen, 2003).

Radio-adaptation has also been reported in normal cells other than those of the immune system. Broome and colleagues demonstrated that exposure of cultured AG1522 human fibroblasts to total doses ranging from 0.0001 Gy to 0.5 Gy at dose rates ranging from 0.001 to 0.003 Gy/min prior to a challenge dose of 4 Gy g -rays exhibited reduced frequency of micronuclei (Broome et al, 2002). There was no significant difference in the degree of protection induced by the two extremes of total doses used and a similar degree of protection was induced with protracted exposure to g -rays and ³H beta particles. These latter findings demonstrated that doses as low as one track per cell (0.0001 Gy) produced the same maximum adaptive response as did doses that resulted in many tracks per cell (0.5 Gy) and that the two types of radiations were similar in this respect. In addition, induction of radio-adaptation in the fibroblasts occurred even when the cells were incubated at 0^o C during delivery of the priming dose, a temperature at which DNA repair and other metabolic processes are inactive.

In a study of human erythrocytes subjected to g -radiation, identical total doses delivered in a single fraction or split into two identical fractions with a 3.5 hr interval between exposures, resulted in a 2.4-fold reduction in hemolysis when the split doses were administered (Koziczak et al, 2003). The results also suggested that the reduced damage to the red blood cells occurred because of a decrease in the level of damage to membrane lipids. An interesting study of genetic damage in erythrocytes in the bone marrow of SHK mice has been recently performed (Zaichkina et al, 2003). The effect of low-dose g -radiation (0.1 and 0.2 Gy total doses at 0.125 Gy/min) on high-dose (1.5 Gy total dose at 1 Gy/min) radiation-induced and spontaneous levels of cytogenetic damage was evaluated over the entire lifetime of the animals. The amount of micronucleated polychromatic erythrocytes in primed, primed and challenged and control groups was assessed at various time points. A single low-dose exposure induced cytogenetic radio-adaptation at 1, 3, 6, 9 and 12 months after priming, regardless of mouse age at the time of priming irradiation. In addition, the low-dose priming exposure resulted in decreases in cytogenetic damage to a level below the spontaneous rate at the end of the lifetime (20 months) of the animals. These data demonstrated that the mechanisms underlying radio-adaptation in erythrocytes protect against chromosome damage induced by high-dose irradiation and appear to also minimize spontaneous mutagenesis during aging.

Although the mechanisms underlying radio-adaptive responses remain unclear, a possible scenario has been proposed by Ikushima and colleagues (1996.) and reviewed by Bhattacharjee and Ito (2001). In this model, low dose pre-irradiation damages a small amount of nuclear DNA, ultimately leading to enhanced capacity to repair serious DNA damage induced by a subsequent high dose irradiation event. In support of this possibility are recent studies demonstrating that continuous exposure to X-rays at a low-dose rate leads to no detectable misrejoining of double-strand DNA breaks in cultured mammalian cells (Kuhne et al, 2002). This finding suggests that the probability of inappropriate repair decreases dramatically when the breaks are separated in time and space. Using cell lines deficient in non-homologous end-joining, the investigators concluded that this is an efficient pathway for correct rejoining of separated broken ends, but that it generates genomic rearrangements if the breaks are close in time and space. In this same study there was no significant decrease in double-strand DNA misrejoining after exposure of the cells to fractionated doses of alpha particles, indicating that radiation quality may be a significant factor in determining whether or not protracted low-dose radiation results in protection.

However, the overall evidence for more efficient DNA repair in radio-adaptation is largely indirect and efforts continue to identify other mechanisms. For example, Takahashi et al, reported that splenocytes from C57BL/6N mice primed with whole-body irradiation to a total dose of 1.5 Gy g -rays delivered at a dose rate of 0.001 Gy/min over 25 hr, exhibited significant suppression in p53, bax and apoptosis after a 3 Gy challenge dose of X-rays delivered at a high-dose rate (1 Gy/min) (Takahashi et al, 2001). In subsequent studies, these same investigators demonstrated that the apoptosis induced by acute 3 Gy irradiation was significantly suppressed in the splenic white pulp of wild-type, but not SCID (severe combined immunodeficiency), mice and that DNA-dependent protein kinase activity may play a major role in the radio-adaptive response following pre-irradiation at low doses (0.15 - 0.6 Gy) (Takahashi et al, 2002, 2003). Others have reported that protein kinase C-mediated signaling may be a key step for transducing the low dose-induced protective signal (Rigaud and Moustacchi, 1996). Using chromosomal aberrations as the indicator of damage in human lymphocytes, it has been demonstrated that the anti-mutagenic action observed with low-dose irradiation is similar to that seen with interferon pre-treatment (Tskhovrebova et al, 1995; Makedonov et al, 2000). These results suggest that priming with low-dose radiation and the action of IFN-g may induce common pathways of protection. Some reports indicate that reactive oxygen species play an important role in inducing radioprotection. In the body, oxygen radicals are generated not only by the direct effects of ionizing radiation, but also by inflammatory cells that migrate to the site of tissue damage (Gridley et al, 2005). Feinendegen has recently proposed that the site, type and size of oxygen radical bursts, as well as the time interval between them, may be an integral component of the mechanisms leading to cellular radio-adaptation (Feinendegen, 2002). Reactive oxygen species at low concentrations appear to induce cellular protection, whereas the reverse is true at high concentrations. In a study of bystander effects, Iyer and Lehnert transferred supernatants from human lung fibroblasts (HFL-1) irradiated with a 0.01 Gy dose of g -rays to non-irradiated HFL-1 cells (Iyer and Lehnert, 2002). When the unirradiated cells were subjected to 2 and 4 Gy doses of acute g -radiation, clonogenic survival was enhanced. The radio-adaptive bystander effect was preceded by an increase in intracellular oxygen radicals, an increase in the redox and DNA repair protein AP-endonuclease and a decrease in p53 protein. Studies with plants suggest that exposure to very low-dose-rate g -rays (66 mSv/hr or approximately 66 mGy/hr) leads to an efficient induction of superoxide dismutase and other enzymes that protect against reactive oxygen species (Zaka et al, 2002).

The effects of long-term, low-level g -irradiation among individuals exposed in radiocontaminated buildings has also been studied. Increased frequencies of various chromosomal aberrations, especially breakpoints, inversions and translocations on chromosomes 7 and 14, have been reported (Hsieh et al, 2002). This is of interest, since the genes that code for the T cell antigen receptor and the heavy chains of antibodies are found on these two chromosomes. Thus, mutations in these chromosomes could lead to significant immunodeficiency.

Some additional findings indicate that doses as low as 0.05 Gy to 1 Gy increased the incidence of cardiovascular disease in both A-bomb survivors and victims of the Chernobyl accident (reviewed by Trivedi and Hannan, 2004). Furthermore, wounds heal more slowly under conditions of radiation exposure, with the delay in wound healing being related directly to the dose of radiation (Ran et al, 2003; Shi et al, 2003). During wartime, large numbers of military personnel may experience prolonged exposure to radiation that occurs simultaneously with wounding. Astronauts are inevitably exposed to chronic irradiation and may also become injured while on board a spacecraft. Combined injury also occurs when surgery and radiation are used for cancer therapy. The mechanisms by which radiation delays wound healing are not clear, but may include reduction in the number of platelets and inflammatory and other tissue-repairing cells that accumulate at the injured site (Qu et al, 2003).

It is often stated that the major goal of cancer radiotherapy is to select a treatment regimen that has the highest probability for uncomplicated cure. A large amount of data suggests that differences in radiosensitivity between tumor and normal cell populations could be exploited to achieve a substantial increase in the therapeutic ratio. The overall implication of the discussed studies is that increased control of at least some types of cancers may be achieved by combining protracted irradiation with standard external beam therapy. However, at this point in time, it is difficult to predict with a high degree of certainty the tumor types for which this treatment approach may be beneficial and those for which the strategy may be a poor choice.

Induction of hyper-radiosensitivity in tumor cells by protracted radiation exposure is clearly not a universal phenomenon. In addition, there appears to be lack of a consistent trend in the extent that tumor cells repair sublethal radiation-induced damage compared to normal cells, although a correlation between the degree of sublethal damage repair and radioresistance does appear to exist. Using a series of assumptions and mathematical modeling, Hahnfeldt and Hlatky have demonstrated that there is a tendency toward an inverse dose-rate effect even when heterogenous and cycling cell populations are exposed to low-LET radiations (e.g., photons and protons) in a protracted manner (Hahnfeldt and Hlatky, 1998). However, the mechanisms by which protracted irradiation increases tumor cell radiosensitivity in relation to acute iurradiation remain elusive. A recent publication by Marples and colleagues elegantly summarizes the most pertinent literature in this regard (Marples et al, 2004). The paper also proposes a three-component model consisting of damage recognition, signal transduction and damage repair. Undoubtedly, many questions still remain unanswered. Especially intriguing is why some tumor cells fail to exhibit hyper-radiosensitivity and why there is inconsistency between accumulation in the radiosensitive G2 phase of the cell cycle and hyper-radiosensitivity. The mechanisms underlying the radio-adaptive response and the significant anti-tumor effect following whole-body irradiation that exceeds what is seen when the same dose is delivered only to the tumor are phenomena that also warrant further investigation. From the information currently available, it appears likely that hyper-radiosensitivity and adaptation leading to increased radioresistance are distinct examples of 'active' responses to low-dose radiation exposure that are associated, at least partly, with the rate of DNA damage and the extent of DNA repair (Joiner et al, 1999; Marples et al. 1997). With the rapid progress currently on-going in elucidating radiation-induced chromosomal events, genes that control DNA repair and signal transduction pathways, incorporation of protracted low-dose radiation in cancer therapy may eventually become common in the clinic.

We suggest that the data, when taken in toto, indicate that future application of reduced dose-rates in the clinic may have several important roles: 1) direct tumor cell cytotoxicity that is greater than what is observed with acute irradiation; 2) induced changes that result in greater tumor control by HDR external-beam radiation; 3) radio-adaptation in normal cell populations that decrease risk for complications; and 4) up-regulation of anti-tumor immune responses. Increased knowledge of mechanistic effects of reduced dose-rate irradiation on both tumor and normal tissues and the subsequent impact on the therapeutic ratio is needed for future clinical utility.

Finally, we accept that the database that we have reviewed is complex and currently it is difficult to proceed to clinical trials with confidence in the application of low dose-rate therapy. However, we do suggest that there are laboratory and clinical research projects that should be undertaken with some priority. First, we need increased knowledge of mechanistic effects of reduced dose-rate irradiation on tumor cells, especially defining the molecular processes that are shared by induced hypersensitivity, adaptive response, and their contribution to cell killing by mitotic and apoptotic processes. Further, it is important that these observations be extended to model tumor systems such as xenografts growing in the nude mouse to determine whether cell-based observations in vitro can be extended to in vivo response of model tumors comprised of cells that do or do not exhibit low dose, low dose-rate radiosensitivity. We think that certain clinical studies should be considered. First, the combination of low dose-rate irradiation delivered by brachytherapy and radiolabelled antibody therapy with external beam high dose-rate radiotherapy is immediately feasible. Second, as soon as variation of response to low dose-rate irradiation and to combined LDR and HDR of brain tumors can be predicted for such tumors with some confidence, then such therapy should be given priority. Also, radioresistant tumors such as certain sarcomas, when located in sites accessible to direct injection of carriers of radionuclides could be considered for LDR/HDR therapy. Last, but not least, a better understanding of the response of normal tissues and the subsequent impact on the therapeutic ratio is needed for future clinical utility.

We thank Larry E. Dillehay, Ph.D., Research Associate and Director of Experimental Radiators in the Department of Radiation Oncology & Molecular Biology at Johns Hopkins School of Medicine for use of data depicted in the figure and William Preston, Ed.D. for expert editorial assistance.