INTRODUCTION

The phenomenon of low-dose hyper-radiosensitivity (HRS), an effect in which cells die from excessive sensitivity to low doses (<0.5 Gy) of ionizing radiation but become more resistant (IRR) to larger doses, has been detected in about 80% of the human tumor cell lines assessed so far (1–4). The suggested mechanism for HRS is related to the absence at low doses of an inducible DNA repair mechanism observed at higher doses, above a putative damage threshold. Therefore, cells may show hypersensitivity to X-ray doses that produce damage that is insufficient to activate this process (5, 6).

Since HRS was reported after fractionated X irradiation in tumor cell lines, indicating recoverability of HRS between fractions (5, 7), there has been a considerable interest in exploiting the HRS phenomenon in radiotherapy of cancer patients. It was suggested that radioresistant (to conventional doses) tumors, such as glioma, could be cured more effectively with multiple very small doses per fraction—“ultrafractionation”—than with conventional radiotherapy (4, 7). In a clinical study on the HRS effect in human tumors, Harney et al. (8) found that an “ultrafractionated” radiotherapy (0.5 Gy three times a day) was more effective in growth delay of metastatic melanoma and sarcoma tumor nodules than conventionally fractionated radiotherapy (1.5 Gy/day). If the HRS phenomenon also exists in normal tissues, one of the consequences of greater effectiveness of ultrafractionation could be more severe side effects. This can be important, especially when intensity-modulated radiotherapy is used, because a larger volume of normal tissue receives very small doses per fraction. Therefore, to obtain therapeutic gain with ultrafractionation, critical normal tissues must be less sensitive to low doses than the tumors.

Hyper-radiosensitivity in normal human cells was first discovered in vitro in experiments measuring the survival of lung epithelial cells after single doses of X rays (9). Three clinical studies of HRS in normal human tissues were focused on the effects of low doses per fraction on skin (4, 10, 11). In two of them, the basal cell density in human epidermis was used as the end point. Turesson et al. (4) found a significant reverse fractionation effect (HRS effect), with greater loss of basal cells after 0.45 Gy than after 1.1 Gy per fraction. However, the effect was lost when dose intensity was taken into account (4, 11). To avoid the impact of the time factor, Harney et al. (11) compared the effects of two regimens of equal dose intensity (0.5 Gy three times a day × 12 days compared to 1.5 Gy/day × 12 days) and did not observe HRS in seven of eight patients after doses of ∼0.5 Gy. In one patient, however, an HRS effect, seen as a significant reduction in basal cell density after low doses, was found, suggesting that HRS may be an individual characteristic. The concept of interindividual variation in normal tissue radiosensitivity to conventional (>1 Gy) doses is commonly known and was shown in many studies searching for methods that predict normal tissue responses to radiotherapy in cancer patients.

The data on HRS in vitro in normal cells are scarce (4, 9). No studies have tested the HRS effect in vitro in primary normal cells from a group of cancer patients. Therefore, the aim of the present study was to examine, using the micronucleus (MN) assay, the low-dose radiation response of normal skin cells (fibroblasts and keratinocytes) derived from cancer patients and to determine whether an HRS-like phenomenon occurs in cells of these patients. If low-dose chromosomal hypersensitivity is present in vitro in the cells of these patients, the future observation of their normal tissue reactions after radiotherapy could address the clinical relevance of this effect.

MATERIALS AND METHODS

Statistical Analysis

The MN data for each patient were fitted to the linear-quadratic (LQ) model (Eq. 1). To study the possible low-dose hypersensitivity (deviation from the LQ curve), for each patient, the induced-repair (IR) model (Eq. 2) was also fitted over the whole data set (0.05–4 Gy).

The IR model, originally suggested by Joiner and Johns (14) to better describe the low-dose response, is based on the LQ model with a modification of the α component, where d is dose, α_r is α extrapolated from the high (conventional)-dose response, and α_s is α derived from the response at very low doses. α_s > α_r represents increased sensitivity at very low doses. d_c is a parameter describing the range of doses over which the transition from hypersensitivity to induced resistance occurs (when α_s to α_r is 63% complete). The β term from the LQ equation remains unmodified. The data (MN induction) were fitted with the IR model and LQ model using nonlinear least-squares regression using the iterative method of Gauss-Newton (Statistica 6.0) to produce the best-fit parameters for each model. The presence of the HRS/IRR dose response is supported by values of α_s higher than α_r, the confidence limits of which do not overlap, and values of d_c significantly greater than 0 (3).

RESULTS

The raw data obtained in the dose–response study for fibroblasts and keratinocytes of each patient are summarized in Tables 1 and 2, respectively. Data are shown for the number of binucleated cells with micronuclei only. However, mean values (±SEM) of percentages of binucleated cells, fractions of binucleated cells with micronuclei, and the numbers of micronuclei per single binucleated cell for the patients' cells are listed in Table 3. Since, in the present study, exactly the same results for two MN induction parameters (fraction of binucleated cells with micronuclei and micronuclei/binucleated cells) were observed, for clarity, the results for only one (fraction of binucleated cells with micronuclei) are presented.

Chromosomal radiosensitivities of fibroblasts as well as keratinocytes, expressed as the induction of micronuclei per unit dose (fraction of binucleated cells with micronuclei/ Gy), varied significantly between cancer patients (P < 0.0005). The high interindividual variation in MN induction was observed after low (<0.5 Gy) and higher doses (Tables 1, 2). For example, there was a sixfold difference at 0.25 Gy between the least responsive and the most responsive patients. As shown in Table 1, there was no relationship between the age of the patients and MN induction at low and high doses.

A linear-quadratic dose response for MN induction was observed for most of the patients, and the dose response approximated linearity for only a few patients. To study the possible low-dose hypersensitivity (HRS), for each patient, the IR model was fitted to the complete data set covering the whole dose range, from 0.05 to 4 Gy. In the fits of the IR model, the deviation over the LQ curve for MN induction after low doses, which is an indicative of an HRS-like response, was demonstrated for the fibroblasts of two patients (Fig. 1) and for the keratinocytes of four other patients (Fig. 2). Values of parameters of the two models for the six patients are shown in Table 4. Values of α_s (derived from the response at very low doses) higher than α_r (derived from the response at high doses) whose confidence limits do not overlap were found for them. The interindividual variability of parameter α_s was higher than of parameter α_r, and it gave a difference between the six patients in the α_s/α_r ratio that ranged from 2.7 to 15.4. The values of d_c were significantly greater than 0 and ranged from 0.13 to 0.36 Gy. For the patients without a low-dose hypersensitive response, the α_s/α_r ratio was ∼1, and the fits of two models were equivalent in the low- and high-dose regions (data not shown).

When the mean MN induction per unit dose (for the whole group of patients) was compared for low and high γ-ray doses, it appeared that doses in the range 0.05–0.25 Gy were significantly (P < 0.0001) more effective per gray than higher doses in both fibroblasts and keratinocytes (Fig. 3). For fibroblasts, the mean effect per gray increased by a factor of ∼1.5 from 0.13 at a dose of 1 Gy to 0.19 at a dose of 0.05 Gy, whereas in keratinocytes the mean effect per gray increased by a factor of ∼3 from 0.05 at a dose of 1 Gy to 0.15 at a dose of 0.1 Gy. If the data were reanalyzed excluding the six patients presenting an HRS-like response, the mean MN induction per unit dose was still significantly higher (∼1.5 times) at doses <0.5 Gy than at higher doses in fibroblasts and keratinocytes (P < 0.0001). The doses in the range 0.75–4 Gy were equally effective per unit dose in MN induction.

Comparison of the means of MN induction for fibroblasts and keratinocytes obtained after irradiation showed statistically significant differences between them (P < 0.0001). In general, keratinocytes were at least twice as resistant to MN induction as fibroblasts (Table 3).

The analysis of 35 patients showed no significant relationship between the chromosomal radiosensitivity of fibroblasts and keratinocytes from the same donor in the low-dose (0.1–0.25 Gy) region of the radiation response (r = 0.02, P = 0.84). The correlation was observed in the high-dose (0.5–4 Gy) region only (r = 0.82, P = 0.0001, Fig. 4). The same result was encountered when the relationship was analyzed for six patients presenting a low-dose HRS-like response. For fibroblasts of two such patients, the calculated correlation coefficients were r = 0.98, P = 0.0001 and r = 0.05, P = 0.93 for high and low doses, respectively, whereas for keratinocytes of the remaining four patients, the values were r = 0.88, P = 0.0001 and r = −0.33, P = 0.25 for high and low doses, respectively.

DISCUSSION

The present study showed high interindividual variation in low-dose (<0.5 Gy) chromosomal radiosensitivity, expressed as MN induction, in a group of patients with cervix cancer. This gave us a reason to investigate whether an HRS-like phenomenon occurs in normal cells of these patients. The data presented here suggest that low-dose chromosomal hypersensitivity is not a common finding in human normal cells and can represent an individual characteristic. An HRS-like response after low doses (seen as the deviation over the LQ curve) was demonstrated only for the fibroblasts of two patients (Fig. 1) and for the keratinocytes of four other patients (Fig. 2). Although an increased effectiveness of low doses (<0.5 Gy) in comparison with higher doses in mean MN induction per unit dose was observed for the whole group of patients, the fibroblasts of 38 patients and keratinocytes of 31 patients did not demonstrate any HRS-like effect.

In the present study we used the MN assay, a simple chromosomal radiosensitivity test with the end point being the result of DNA damage and repair. The usefulness of the MN assay as a biological indicator of radiation doses as low as 0.1 Gy was confirmed in many studies (15–17). These results show that it is possible to observe the deviation over the LQ curve for MN induction after low doses, which is indicative of an HRS-like response. Similarly, in the study of Courdi et al. (16), the MN assay allowed them to observe significant chromosomal hypersensitivity to low doses in some of the human tumor cell lines they examined.

For reasons given below, we cannot exclude the possibility that the HRS response would be seen in more patients if different culture conditions and different radiosensitivity assays were used. It was shown recently that low-dose hyper-radiosensitivity is dependent on the cell cycle and is a consequence of ineffective cell cycle arrest of cells irradiated in G₂ phase with doses less than 0.4 Gy (18–20). Short et al. (18) found that in some cell lines HRS was observed in asynchronous populations but always with the most marked effect in G₂-phase cells. In some cell lines, however, HRS was not found in asynchronous populations, although it was demonstrable in G₂-phase cells (18, 20). Therefore, it is possible that at least in some of the patient cells in our study, the HRS response at G₂ phase could be masked in the asynchronous cell cultures we used. Another less likely possibility is that the cells of some patients demonstrate low-dose hypersensitivity, but only at much lower doses than it is possible to study with the MN assay. An increased biological effectiveness per unit dose (1.5 times) of the lowest doses, 0.05 Gy for fibroblasts and 0.1 Gy for keratinocytes in comparison with higher doses, was still observed for the group of patients when the six patients with an HRS-like response were excluded from the group.

In the present study, high variability in the HRS-like response between six patients was observed. The α_s/α_r ratio ranged from 2.7 to 15.4; the difference was due to the high interindividual variation in the values of α_s representing the response at very low doses (Table 4). The parameter d_c in the IR model describes the range of doses over which the transition from HRS to IRR occurs. The d_c values obtained in the present study for six patients with HRS ranged from 0.13 to 0.36 Gy. Therefore, the difference in both parameters between the patients suggests that the HRS phenomenon can be an individual characteristic. The present d_c values are similar to those obtained with the clonogenic assay by other authors (3) for cells of human tumor cell lines (0.14–0.47 Gy) but lower than the value of d_c reported by Singh et al. (9) for normal human lung epithelial cells (0.59 Gy). Only one report on MN induction in tumor cells after low-dose irradiation is known in which the authors found an HRS response in one tumor cell line (16), but they did not use the IR model to fit the data, and the d_c value is not known.

Comparison of the means of MN induction for fibroblasts and keratinocytes obtained after irradiation has shown keratinocytes to be at least twice as resistant as fibroblasts (Table 3). In the previous work (13), greater MN induction in human fibroblasts than in keratinocytes was found after X irradiation. Similarly, in cell survival studies, Geara et al. (21) and D'Errico et al. (22) found keratinocytes to be more radioresistant (higher surviving fraction) than fibroblasts after irradiation.

The difference in radiosensitivity between these two types of cells can be one explanation of the more marked HRS-like effect seen in keratinocytes (higher α_s/α_r ratio ranging from 4–15.4) than in fibroblasts (α_s/α_r ratio ranging from 2.7–7.9). This is consistent with the previous suggestion that the low-dose effect is more pronounced in more radioresistant cells (3). In the cell survival studies of Singh et al. (9) and Short et al. (3), an α_s/α_r ratio of 7.9 for normal human lung epithelial cells and an α_s/α_r ratio ranging from 13.6 to 21.6 for the most radioresistant tumor cell lines have been reported. In regard to the greatest HRS effect seen in G₂ phase, it was suggested that actively proliferating cell populations may demonstrate a greater increase in radiosensitivity to very low doses compared with quiescent populations (18). Therefore, quickly proliferating keratinocytes in comparison with slowly proliferating fibroblasts could have more chance to show the HRS effect.

We observed no correlation between the chromosomal radiosensitivity of fibroblasts and keratinocytes after very low doses (0.1–0.25 Gy), although a positive correlation was seen after higher doses (0.5–4 Gy) (Fig. 4). Therefore, the results seen after low doses suggest that the response to DNA damage can be cell type specific. Flatt et al. (23) and D'Errico et al. (22) have reported that there is a difference in the cell cycle response for DNA damage between human keratinocytes and fibroblasts. The G₁ arrest in keratinocytes was attenuated, whereas fibroblasts were arrested in G₁ phase after γ irradiation. Second, our results suggest that cell type specificity for DNA damage can be dependent on the radiation dose. Ding et al. (24) recently found differences in gene expression of human fibroblasts after low-dose (2 cGy) and high-dose (4 Gy) irradiation. At low dose, genes involved in cell-cell signaling, signal transduction and response to DNA damage were predominantly expressed, whereas at high dose, genes involved in apoptosis and cell proliferation were predominantly expressed. In the study by D'Errico et al. (22), apoptosis was not observed after high-dose X irradiation in normal human keratinocytes and fibroblasts. Therefore, the difference between the two types of cells after higher doses (>0.5 Gy) can be diminished and give a positive correlation between them. Because some genes were found to respond only after low doses (24), all this suggests that cell type specificity for DNA damage can be manifested more easily at low doses. This could explain, at least in part, why in our study the low-dose hypersensitive response demonstrated in one type of cells was not seen in another type of cells from the same donors.

In conclusion, our study showed that fibroblasts of two and keratinocytes of four of the 40 patients studied are hypersensitive to radiation doses <0.4 Gy and show induced radioresistance in response to higher doses. The data suggest that low-dose chromosomal hypersensitivity is not a common finding in normal human cells and can be an individual characteristic. The existence of HRS in normal cells of some patients can imply, on the one hand, lower cancer risk from small exposures to ionizing radiation, because the elimination of damaged cells protects them from mutation. The first experimental evidence for that was provided by Redpath et al. (20), who observed the reduction of transformation frequency at low doses as a consequence of low-dose HRS. On the other hand, the HRS effect can be responsible for more severe normal tissue response to radiotherapy. The clinical relevance of the low-dose chromosomal hypersensitivity in the normal cells of the six patients is under investigation after their course of radiotherapy.