Low-dose radiation response of primary keratinocytes and fibroblasts from patients with cervix cancer

The aim of the present study was to examine, using the micronucleus (MN) assay, the low-dose radiation response of normal skin cells from cancer patients and to determine whether the hyper-radiosensitivity (HRS)-like phenomenon occurs in cells of these patients. Primary skin fibroblasts and keratinocytes derived from 40 patients with cervix cancer were studied. After in vitro gamma irradiation with single doses ranging from 0.05 to 4 Gy, MN induction was assessed. For each patient, the linear-quadratic (LQ) model and the induced repair (IR) model were fitted over the whole data set. In fits of the IR model, an HRS-like response after low doses (seen as the deviation over the LQ curve) was demonstrated for the fibroblasts of two patients and for the keratinocytes of four other patients. The alpha(s)/alpha(r) ratio for the six patients ranged from 2.7 to 15.4, whereas the values of the parameter d(c) ranged from 0.13 to 0.36 Gy. No relationship was observed between chromosomal radiosensitivity of fibroblasts and keratinocytes derived from the same donor in the low-dose (0.1-0.25 Gy) region. In conclusion, the fact that low-dose chromosomal hypersensitivity was observed for cells of only six of the patients studied suggests that it is not a common finding in human normal cells and can represent an individual characteristic.     

Effects of exposure to low-dose-rate (60)co gamma rays on human tumor cells in vitro

Cells of three asynchronously growing human tumor cell lines, PC3 (human prostate carcinoma), T98G and A7 (human glioblastomas), which have been shown previously to demonstrate low-dose hyper-radiosensitivity to low acute single doses, were irradiated with (60)Co gamma rays at low dose rates (2 cGy-1 Gy h(-1)). Instead of a dose-rate sparing response, these cell lines demonstrated an inverse dose-rate effect on cell survival at dose rates below 1 Gy h(-1), whereby a decrease in dose rate resulted in an increase in cell killing per unit dose. A hyper-radiosensitivity-negative cell line, U373MG, did not demonstrate an inverse dose-rate effect. Analysis of the cell cycle indicated that this inverse dose-rate effect was not due to accumulation of cells in G(2)/M phase or to other cell cycle perturbations. T98G cells in reversible G(1)-phase arrest also showed an inverse dose-rate effect at dose rates below 30 cGy h(-1) but a sparing effect as the dose rate was reduced from 60 to 30 cGy h(-1). We conclude that this inverse dose-rate effect in continuous exposures reflects the hyper-radiosensitivity seen in the same cell lines in response to very small acute single doses.     

An association between low-dose hyper-radiosensitivity and the early G2-phase checkpoint in normal fibroblasts of cancer patients

In our previous study, low-dose hyper-radiosensitivity (HRS) effect was demonstrated for normal fibroblasts (asynchronous and G2-phase enriched) of 4 of the 25 cancer patients investigated. For the rest of patients, HRS was not defined in either of the 2 fibroblast populations. Thus, the study indicated that G2-phase enrichment had no influence on HRS identification. The conclusion contradicts that reported for human tumor cells, and suggests different mechanism of HRS in normal human cells. In the present paper we report, for the first time, the activity of early G2-phase checkpoint after low-dose irradiation in normal fibroblasts of these 4 HRS-positive patients and 4 HRS-negative patients and answer the question regarding the role of this checkpoint in normal human cells. The response of the early G2-phase checkpoint was determined by assessment of the progression of irradiated cells into mitosis using the mitotic marker, phosphorylated histone H3. We found evident differences in the activity of the early G2-phase checkpoint between HRS-positive and HRS-negative fibroblasts. In HRS-positive fibroblasts the checkpoint was not triggered and DNA damage was not recognized after doses lower than 0.2 Gy resulting in HRS response. On the contrary, in HRS-negative fibroblasts the early G2-phase checkpoint was activated regardless of the dose in the range 0.1–2 Gy. In conclusion, although cell cycle phase has no effect on the presence of HRS effect in normal human fibroblasts, the data reported here indicate that HRS response in these cells is associated with the functioning of early G2-phase checkpoint in a threshold-dose dependent manner, similarly as it takes place in most of human tumor and other cells.     

Equivalence of pulsed-dose-rate to low-dose-rate irradiation in tumor and normal cell lines

To determine whether different fractionation schemes could simulate low-dose-rate irradiation, ovarian cells of the carcinoma cell lines A2780s (radiosensitive) and A2780cp (radioresistant) and AG1522 normal human fibroblasts were irradiated in vitro using different fraction sizes and intervals between fractions with an overall average dose rate of 0.53 Gy/h. For the resistant cell line, the three fractionation schemes, 0.53 Gy given every hour, 1.1 Gy every 2 h, and 1.6 Gy every 3 h, were equivalent to low dose rate (0.53 Gy/h). Two larger fraction sizes, 2.1 Gy every 4 h and 3.2 Gy every 6 h, resulted in lower survival than that after low-dose-rate irradiation for the resistant cell line, suggesting incomplete repair of radiation damage due to the larger fraction sizes. The survival for the sensitive cell line was lower at small doses, but then it increased until it was equivalent to that after low-dose-rate irradiation for some fractionation schemes. The sensitive cell line showed equivalence only with the 1.6-Gy fraction every 3 h, although 0.53 Gy every 1 h and 1.1 Gy every 2 h showed equivalence at lower doses. This cell line also showed an adaptive response. The normal cell line showed a sensitization to the pulsed-dose-rate schemes compared to low-dose-rate irradiation. These data indicate that the response to pulsed-dose-rate irradiation is dependent on the cell line and that compared to the response to low-dose-rate irradiation, it shows some equivalence with the resistant carcinoma cell line, an adaptive response with the parental carcinoma cell line, and sensitization with the normal cells. Therefore, further evaluation is required before implementing pulsed-dose-rate irradiation in the clinic.     

Induction of micronuclei in human fibroblasts and keratinocytes by 25 kV x-rays

A relative biological effectiveness (RBE) not much larger than unity is usually assumed for soft x-rays (up to approximately 50 keV) that are applied in diagnostic radiology such as mammography, in conventional radiotherapy and in novel radiotherapy approaches such as x-ray phototherapy. On the other hand, there have been recent claims of an RBE of more than 3 for mammography and respective conventional x-rays. Detailed data on the RBE of soft x-rays, however, are scarce. The aim of the present study was to determine the effect of low-energy x-rays on chromosomal damage in vitro, in terms of micronucleus induction. Experiments were performed with 25 kV x-rays and a 200 kV x-ray reference source. The studies were carried out on primary human epidermal keratinocytes (HEKn), human fibroblasts (HFIB) and NIH/3T3 mouse fibroblasts. Micronucleus (MN) induction was assayed after in vitro irradiation with doses ranging from 1 to 5.2 Gy. Compared to the effect of 200 kV x-rays, 25 kV x-rays resulted in moderately increased chromosomal damage in all cell lines studied. This increase was observed for the percentage of binucleated (BN) cells with micronuclei as well as for the number of micronuclei per BN cell. Moreover, the increased number of micronuclei per micronucleated BN cell in human keratinocytes and 3T3 mouse fibroblasts suggests that soft x-rays induce a different quality of damage. For all cell lines studied the analysis of micronucleus induction by 25 kV soft x-rays compared to 200 kV x-rays resulted in an RBE value of about 1.3. This indicates a somewhat enhanced potential of soft x-rays for induction of genetic effects.     

Critical target and dose and dose-rate responses for the induction of chromosomal instability by ionizing radiation.

To investigate the critical target, dose response and dose-rate response for the induction of chromosomal instability by ionizing radiation, bromodeoxyuridine (BrdU)-substituted and unsubstituted GM10115 cells were exposed to a range of doses (0.1-10 Gy) and different dose rates (0.092-17.45 Gy min(-1)). The status of chromosomal stability was determined by fluorescence in situ hybridization approximately 20 generations after irradiation in clonal populations derived from single progenitor cells surviving acute exposure. Overall, nearly 700 individual clones representing over 140,000 metaphases were analyzed. In cells unsubstituted with BrdU, a dose response was found, where the probability of observing delayed chromosomal instability in any given clone was 3% per gray of X rays. For cells substituted with 25-66% BrdU, however, a dose response was observed only at low doses (<1.0 Gy); at higher doses (>1.0 Gy), the incidence of chromosomal instability leveled off. There was an increase in the frequency and complexity of chromosomal instability per unit dose compared to cells unsubstituted with BrdU. The frequency of chromosomal instability appeared to saturate around approximately 30%, an effect which occurred at much lower doses in the presence of BrdU. Changing the gamma-ray dose rate by a factor of 190 (0.092 to 17.45 Gy min(-1)) produced no significant differences in the frequency of chromosomal instability. The enhancement of chromosomal instability promoted by the presence of the BrdU argues that DNA comprises at least one of the critical targets important for the induction of this end point of genomic instability.     

Bystander and delayed effects after fractionated radiation exposure

Human immortalized keratinocytes were exposed to a range of single or fractionated doses of gamma rays from (60)Co, to medium harvested from donor cells exposed to these protocols, or to a combination of radiation and irradiated cell conditioned medium (ICCM). The surviving fractions after direct irradiation or exposure to ICCM were determined using a clonogenic assay. The results show that medium harvested from cultures receiving fractionated irradiation gave lower "recovery factors" than direct fractionated irradiation, where normal split-dose recovery occurred. The recovery factor is defined here as the surviving fraction of the cells receiving two doses (direct or ICCM) separated by an interval of 2 h divided by the surviving fraction of cells receiving the same dose in one exposure. After treatment with ICCM, the recovery factors were less than 1 over a range of total doses from 5 mGy-5 Gy. Varying the time between doses from 10 min to 180 min did not alter the effect of ICCM, suggesting that two exposures to ICCM are more toxic than one irrespective of the dose used to generate the response. In certain protocols using mixtures of direct irradiation and ICCM, it was possible to eliminate the bystander effect. If bystander factors are produced in vivo, then they may reduce the sparing effect of the dose fractionation.     

Relationship between radiation-induced low-dose hypersensitivity and the bystander effect

Recent advances in our knowledge of the biological effects of low doses of ionizing radiation have shown two unexpected phenomena: a "bystander effect" that can be demonstrated at low doses as a transferable factor(s) causing radiobiological effects in unexposed cells, and low-dose hyper-radiosensitivity and increased radioresistance that can be demonstrated collectively as a change in the dose-effect relationship, occurring around 0.5-1 Gy of low-LET radiation. In both cases, the effect of very low doses is greater than would be predicted by conventional DNA strand break/repair-based radiobiology. This paper addresses the question of whether the two phenomena have similar or exclusive mechanisms. Cells of 13 cell lines were tested using established protocols for expression of both hyper-radiosensitivity/increased radioresistance and a bystander response. Both were measured using clonogenicity as an end point. The results showed considerable variation in the expression of both phenomena and suggested that cell lines with a large bystander effect do not show hyper-radiosensitivity. The reverse was also true. This inverse relationship was not clearly related to the TP53 status or malignancy of the cell line. There was an indication that cell lines that have a radiation dose-response curve with a wide shoulder show hyper-radiosensitivity/increased radioresistance and no bystander effect. The results may suggest new approaches to understanding the factors that control cell death or the sectoring of survival at low radiation doses.     

Early changes in mouse urinary bladder function following fractionated X irradiation.

Functional changes in the mouse urinary bladder following single-dose or fractionated irradiation were assessed by cystometry, i.e., by measuring the intravesical volume-pressure relationship during transurethral filling. The early response presented as a dose-dependent and transitory decrease in the reservoir function of the organ as defined by the intravesical volume at a filling pressure of 10 or 20 mm Hg, V10 or V20. The quantal dose response used in the present study was a reduction in the individual bladder volume (V10 and V20) by at least 50%. After single doses greater than or equal to 10 Gy, the reaction occurred between Days 7 and 25 with maximum prevalence between Days 7 and 14. The individual duration of the response was 3-9 days. Treatment with single doses and 2, 3, 5, and 10 fractions demonstrated a significant sparing effect with ED50 values of 18.3, 24.9, 26.8, 29.8, and 38.0 Gy, respectively. The linear-quadratic model fitted the data reasonably well when tested according to Tucker (Int. J. Radiat. Oncol. Biol. Phys. 10, 1933-1939, 1984). The alpha/beta ratios estimated with different two-step techniques ranged from 11.1 to 12.4 Gy. Analysis of the data as proposed by Thames et al. (Int. J. Radiat. Biol. 49, 999-1009, 1986) yielded an alpha/beta value of 13.9 Gy (95% confidence limits 8.4 and 24.6 Gy), illustrating a fractionation effect typical for acutely responding tissues, although no clear cell depletion occurred in the urothelium.     

Low-dose radiation hypersensitivity in human tumor cell lines: effects of cell-cell contact and nutritional deprivation

The hyper-radiosensitivity at low doses recently observed in vitro in a number of cell lines is thought to have important implications for improving tumor radiotherapy. However, cell-cell contact and the cellular environment influence cellular radiosensitivity at higher doses, and they may alter hyper-radiosensitivity in vivo. To confirm this supposition, we investigated the effects of cell density, multiplicity and nutritional deprivation on low-dose hypersensitivity in vitro. Cell survival in the low-dose range (3 cGy to 2 Gy) was studied in cells of two human glioma (BMG-1 and U-87) and two human oral squamous carcinoma (PECA-4451 and PECA-4197) lines using a conventional macrocolony assay. The effects of cell density, multiplicity and nutritional deprivation on hyper-radiosensitivity/induced radioresistance were studied in cells of the BMG-1 cell line, which showed prominent hypersensitivity and induced radioresistance. The induction of growth inhibition, cell cycle delay, micronuclei and apoptosis was also studied at the hyper-radiosensitivity-inducing low doses. Hyper-radiosensitivity/induced radioresistance was evident in the cells of all four cell lines to varying extents, with maximum sensitivity at 10-30 cGy, followed by an increase in survival up to 50 cGy-1 Gy. Both the glioma cell lines had more prominent hyper-radiosensitivity than the two squamous carcinoma cell lines. Low doses inducing maximum hyper-radiosensitivity did not cause significant growth inhibition, micronucleation or apoptosis in BMG-1 cells, but a transient G(1)/S-phase block was evident. Irradiating and incubating BMG-1 cells at high density for 0 or 4 h before plating, as well as irradiating cells as microcolonies, reduced hyper-radiosensitivity significantly, indicating the role of cell-cell contact-mediated processes. Liquid holding of BMG-1 cells in HBSS + 1% serum during and after irradiation for 4 h significantly reduced hyper-radiosensitivity, suggesting that hyper-radiosensitivity may be due partly to active damage fixation processes at low doses. Therefore, our findings suggest that the damage-induced signaling mechanisms influenced by (or mediated through) cell-cell contact or the cellular environment, as well as the lesion fixation processes, play an important role in hyper-radiosensitivity. Further studies are required to determine the exact nature of the damage that triggers these responses as well as for evaluating the potential of low-dose therapy.     

The use of 'top-up' experiments to investigate the effect of very small doses per fraction in mouse skin

The partial tolerance type of 'top-up' experiment has been investigated to determine the resolution of this approach for studying the damage to mouse skin from very small doses of X-rays and neutrons. The effect of 20 fractions, each as small as 0.10 Gy of X-rays or of 0.05 Gy of neutrons, can be detected if 3 MeV neutrons are used as the 'top-up' reference radiation. This capability results from the almost linear underlying dose-response curve and highly reproducible dose-effect relationship for the low energy neutrons. The data fit the linear quadratic model of dose fractionation for X-rays down to fractional doses of 0.75 Gy, but at lower doses there is a trend towards an increase in the skin radiosensitivity. Modelling shows that this might be consistent with a sub-population of the cells showing an exceptional radiosensitivity, and a replenishment of this subpopulation occurring in the 8 h between small dose fractions. More experiments are needed at very low doses in order to confirm this hypothesis for skin and for other tissues.     

Dose- and time-dependent changes in gene expression in human glioma cells after low radiation doses

We have used DNA microarrays to identify changes in gene expression in cells of the radioresistant human glioma cell lines T98G and U373 after low radiation doses (0.2-2 Gy). Using Bayesian linear models, we have identified a set of genes that respond to low doses of radiation; furthermore, a hypothesis-driven approach to data analysis has allowed us to identify groups of genes with defined non-linear dose responses. Specifically, one of the cell lines we have examined (T98G) shows increased radiosensitivity at low doses (low-dose hyper-radiosensitivity, HRS); thus we have also assessed sets of genes whose dose response mirrors this survival pattern. We have also investigated a time course for induction of genes over the period when the DNA damage response is expected to occur. We have validated these data using quantitative PCR and also compared genes up-regulated in array data to genes present in the polysomal RNA fraction after irradiation. Several of the radioresponsive genes that we describe code for proteins that may have an impact on the outcome of irradiation in these cells, including RAS homologues and kinases involved in checkpoint signaling, so understanding their differential regulation may suggest new ways of altering radioresistance. From a clinical perspective these data may also suggest novel targets that are specifically up-regulated in gliomas during radiotherapy treatments.     

Protection from radiation-induced apoptosis by the radioprotector amifostine (WR-2721) is radiation dose dependent

The radioprotective agent amifostine is a free radical scavenger that can protect cells from the damaging effects of ionising radiation when administered prior to radiation exposure. However, amifostine has also been shown to protect cells from chromosomal mutations when administered after radiation exposure. As apoptosis is a common mechanism by which cells with mutations are removed from the cell population, we investigated whether amifostine stimulates apoptosis when administered after radiation exposure. We chose to study a relatively low dose which is the maximum radiation dose for radiation emergency workers (0.25 Gy) and a high dose relevant to radiotherapy exposures (6 Gy). Mice were administered 400 mg/kg amifostine 30 min before, or 3 h after, whole-body irradiation with 0.25 or 6 Gy X-rays and apoptosis was analysed 3 or 7 h later in spleen and bone marrow. We observed a significant increase in radiation-induced apoptosis in the spleen of mice when amifostine was administered before or after 0.25 Gy X-rays. In contrast, when a high dose of radiation was used (6 Gy), amifostine caused a reduction in radiation-induced apoptosis 3 h post-irradiation in spleen and bone marrow similar to previously published studies. This is the first study to investigate the effect of amifostine on radiation-induced apoptosis at a relatively low radiation dose and the first to demonstrate that while amifostine can reduce apoptosis from high doses of radiation, it does not mediate the same effect in response to low-dose exposures. These results suggest that there may be a dose threshold at which amifostine protects from radiation-induced apoptosis and highlight the importance of examining a range of radiation doses and timepoints.     

Dose rate effect on micronuclei induction in human blood lymphocytes exposed to single pulse and multiple pulses of electrons

The effects of single pulses and multiple pulses of 7 MV electrons on micronuclei (MN) induction in cytokinesis-blocked human peripheral blood lymphocytes (PBLs) were investigated over a wide range of dose rates per pulse (instantaneous dose rate). PBLs were exposed to graded doses of 2, 3, 4, 6, and 8 Gy of single electron pulses of varying pulse widths at different dose rates per pulse, ranging from 1 × 10⁶ Gy s⁻¹ to 3.2 × 10⁸ Gy s⁻¹. Different dose rates per pulse were achieved by changing the dose per electron pulse by adjusting the beam current and pulse width. MN yields per unit absorbed dose after irradiation with single electron pulses were compared with those of multiple pulses of electrons. A significant decrease in the MN yield with increasing dose rates per pulse was observed, when dose was delivered by a single electron pulse. However, no reduction in the MN yield was observed when dose was delivered by multiple pulses of electrons. The decrease in the yield at high dose rates per pulse suggests possible radical recombination, which leads to decreased biological damage. Cellular response to the presence of very large numbers of chromosomal breaks may also alter the damage.     

Dose-response relationship of gamma-ray-induced reciprocal translocations at low doses in spermatogonia of the crab-eating monkey (Macaca fascicularis)

The yield of translocations induced by acute gamma-irradiation at low doses (0.25 and 0.50 Gy) in the crab-eating monkey's (Macaca fascicularis) spermatogonia was examined. The frequencies of translocations per cell were 0.53% at 0.25 Gy and 1.07% at 0.50 Gy. Over the low dose range from 0 to 1 Gy, the dose-response relationship for translocation yield was a linear one with a regression coefficient of 1.79 X 10(-2). To estimate the sensitivity to the induction of translocations in the crab-eating monkey's spermatogonia, the slope of the regression line was compared with those in other mammalian species. Consequently, over the low dose range below 1 Gy, the sensitivity of the crab-eating monkey's spermatogonia to translocation induction was similar to several mammalian species, the mouse. Chinese hamster, and the rabbit, but significantly higher than that of the rhesus monkey and lower than that of the marmoset.     

Threshold-like dose of local beta irradiation repeated throughout the life span of mice for induction of skin and bone tumors

The backs of female ICR mice were irradiated with beta rays from 90Sr-90Y three times a week throughout life. Previously we observed 100% tumor incidence at five different dose levels ranging from 1.5 to 11.8 Gy per exposure, but no tumor on repeated irradiation with 1.35 Gy for 300 days (Radiat. Res. 115, 488, 1988). In the present study, delay of tumor development was again seen at a dose of 1.5 Gy per exposure, with further delay at 1.0 Gy. The final tumor incidence was 100% with these two doses. At 0.75 Gy per exposure, no tumor appeared within 790 days after the start of irradiation, but one osteosarcoma and one squamous cell carcinoma did finally appear. These findings indicate a threshold-like response of tumor induction in this repeated irradiation system and further suggest that the apparent threshold may be somewhat less than 0.75 Gy per exposure.     

Effect of thiols on micronucleus frequency in gamma-irradiated mammalian cells

The effects of the thiols cysteamine, WR-1065, and WR-255591 on radiation-induced micronucleus (MN) frequency and cell killing were compared in cultured Chinese hamster ovary cells. MN were measured using the cytochalasin B assay of Fenech and Morley (1985), which minimizes the effect of cytokinetic perturbations on MN expression. The dose-response curves for MN induction were curvilinear both for control cells at doses between 1.5 and 4.5 Gy and for thiol-treated cells at doses between 3 and 9 Gy. Protection against MN induction by each thiol was independent of radiation dose. Furthermore, there was a close correlation between the degree of modification of MN induction and cell survival by each thiol, i.e., the MN frequency closely predicted the survival level regardless of the presence of absence of the thiols. A similar predictive relationship has also been reported by us for cell survival and DNA double-strand break (DSB) induction in this cell line following treatment with these same thiols. Collectively, these data support the hypothesis that, for DNA-repair-proficient mammalian cells treated with radiomodifying agents that do not alter DNA-repair processes, MN and DSB induction are predictive of the level of radiation lethality and of each other.     

On the shape of neutron dose-effect curves for radiogenic cancers and life shortening in mice

Male and female hybrid BCF₁ (C57BL/6 BdxBALB/c Bd) were exposed to total neutron doses of 0.06, 0.12, 0.24, and 0.48 Gy in fractions over a period of 24 weeks. The fractionation regimens were: 24 weekly fractions of 0.0025 Gy, 12 fractions of 0.01 Gy every 2 weeks, 6 fractions of 0.04 Gy every 4 weeks, and 3 fractions of 0.16 Gy every 8 weeks. In order to detect any change in susceptibility with age over the period of exposures from 16 weeks to 40 weeks of age, mice were exposed to single doses of 0.025, 0.05, 0.10, and 0.2 Gy at 16 and 40 weeks of age. These experiments were designed to test whether the initial parts of the dose-response relationships for life shortening and cancer induction could be determined economically by using fractionated exposures and whether or not the initial slopes were linear. The conclusions were that for life shortening and most radiogenic cancers, the dose-effect curves are linear and that fractionation of the neutron dose has no effect on the magnitude of the response of equal total doses over the range of doses studied. The ratio of such initial slopes and comparable linear initial slopes for a reference radiation should provide maximum and constant relative biological effectiveness values.     

The induction of micronuclei in human lymphocytes by low doses of radiation

The appearance of micronuclei (MN) is delayed with respect to cell division in populations of irradiated human lymphocytes, so that the length of time in culture, as well as the number of divisions, is a factor in MN assays. Using two assays that control for cell kinetics, we measured the yield of cells with MN exposed to graded doses of 60Co gamma rays and 90KVP X-rays. The yields showed a non-linear increase with dose. They can be represented by two straight lines: the one in the range below 0.15 Gy has a slight slope, the other in the range above 0.15 Gy has a significantly greater slope. The radical scavengers cysteamine and glycerol, which reduced the MN yields sharply at 3 Gy, were less effective at 0.3 Gy, indicating that terminal deletions arising from the direct ionization of DNA are a major source of the MN induced by low radiation doses. It is likely that the non-linear dose response is due to the saturation of a DNA repair process.     

Renal damage in the mouse: the effect of d(4)-Be neutrons

A further study on the response of the mouse kidney to d(4)-Be neutrons (EN = 2.3 MeV) is described. The results confirm and augment the work published previously by Stewart et al. [Br. J. Radiol. 57, 1009-1021 (1984)]; the present paper includes the data from a "top-up" design of experiment which extends the measurements of neutron RBE (relative to 240 kVp X rays) down to X-ray doses of 0.75 Gy per fraction. The mean RBE for these neutrons increases from 5.8 to 7.3 as X-ray dose per fraction decreases from 3.0 to 1.5 Gy in the kidney. This agrees with the predictions from the linear quadratic (LQ) model, based on the renal response to X-ray doses above 4 Gy per fraction. The mean RBE estimate from a single dose group at 0.75 Gy per fraction of X rays is, however, 3.9. This is below the LQ prediction and may indicate increasing X-ray sensitivity at low doses. Data from this study and from those published previously have been used to determine more accurately the shape of the underlying response to d(4)-Be neutrons; an alpha/beta ratio of 20.5 +/- 3.7 Gy was found. The best value of alpha/beta for X rays determined from these experiments was 3.04 +/- 0.35 Gy, in agreement with previous values.