**Coefficients that describe general radiosensitivity patterns for 39 cell lines**

Fig. 2. Radiosensitivity of five glioblastoma cell lines in relationship to 34 other human tumor cell line data extracted from Williams et al 2008a. The ordinate is the coefficient that describes the slope of the alpha responses and the abscissa is the coefficient that describes the slope of omega responses as shown in figure 1. The diagonal lines are best fit estimates of regression and identify four distinct radiosensitivity groups (VS, S, R and VR). The "VR" (very resistant) group is comprised of only three radioresistant glioblastoma cell lines (U251, T98G, U87). In this figure we two cell lines that are also classified ad glioblastoma cell lines (JW-1T, GL-13) that do not segregate into the VR group. We also identify two R (resistant) cell lines DLD-1 and its subclone abrogated in p21, 19S186.

#### **3.3 Radiosensitivity of radioresistant glioblastoma cell lines and human colorectal tumor cells to low dose-rate irradiation**

We have previously measured the response of 27 human tumor lines to low dose-rate ionizing radiation (26) and in figure 3 we show the general responses of a VR cell line (U-251) and a less resistant colorectal tumor cell line (DLD-1).

These data show two important differences in these two cell lines. First the VR line is more resistant than the R line for both rates of radiation. Second, it is clear that within each line the differences between irradiation at HDR and LDR are markedly different for the two types of cells with the VR line showing a significant increase in inactivation by LDR.

These differences in rates of clonal inactivation between LDR and HDR are shown in more detail in the data in figure 4.

**VS**

**GL-13**

**wt p53 mt p53 null p53**

**Coefficients that describe general radiosensitivity patterns for 39 cell lines** 

**0 0.1 0.2 0.3 0.4**

**S**

**U87**

**19S186 DLD-1**

**omega response**

Fig. 2. Radiosensitivity of five glioblastoma cell lines in relationship to 34 other human tumor cell line data extracted from Williams et al 2008a. The ordinate is the coefficient that describes the slope of the alpha responses and the abscissa is the coefficient that describes the slope of omega responses as shown in figure 1. The diagonal lines are best fit estimates of regression and identify four distinct radiosensitivity groups (VS, S, R and VR). The "VR" (very resistant) group is comprised of only three radioresistant glioblastoma cell lines (U251, T98G, U87). In this figure we two cell lines that are also classified ad glioblastoma cell lines (JW-1T, GL-13) that do not segregate into the VR group. We also identify two R (resistant)

**3.3 Radiosensitivity of radioresistant glioblastoma cell lines and human colorectal** 

We have previously measured the response of 27 human tumor lines to low dose-rate ionizing radiation (26) and in figure 3 we show the general responses of a VR cell line (U-

These data show two important differences in these two cell lines. First the VR line is more resistant than the R line for both rates of radiation. Second, it is clear that within each line the differences between irradiation at HDR and LDR are markedly different for the two

These differences in rates of clonal inactivation between LDR and HDR are shown in more

types of cells with the VR line showing a significant increase in inactivation by LDR.

**0**

**0.05**

**0.1**

**0.15**

**R**

**U251**

cell lines DLD-1 and its subclone abrogated in p21, 19S186.

251) and a less resistant colorectal tumor cell line (DLD-1).

**tumor cells to low dose-rate irradiation** 

detail in the data in figure 4.

**T98G**

**JW-1T**

**VR**

**0.2**

**0.25**

**alpha response**

**0.3**

**0.35**

**0.4**

**0.45**

Fig. 3. Comparison of clonogenic inactivation induced by acute high dose-rate HDR (50 Gy/hr) and protracted irradiation LDR (0.25 Gy/hr). The dashed lines represent the extrapolation of the rate of inactivation at lower doses based on the slopes of inactivation by LDR and HDR.

Radiobiology of Radioresistant Glioblastoma 15

**U251**

**0 40 80 120 160 200 240 (hour)**

**24 hrs omega(HDR)**

**Time**

Fig. 5. Changes in the rate of cell killing expressed as Log10 of cells killed per Gy as a function of time of protracted irradiation in U251 cells irradiated with 4 different constant dose-rates and one that begins at 0.49 Gy/hr and decays with a half life of 2.7 days. The horizontal dotted and dashed lines represent the rate of cell killing for HDR irradiation at

These data show that increasing dose-rate increases the rate at which cells are inactivated until dose-rates reach approximately 0.12 Gy/hr when there is a relatively common response for higher dose-rates including our chosen LDR of 0.25 Gy/hr. The diagonal dashed lines in figure 5 are a general indication of the relative effects of dose-rate compared to the relative effects of duration of exposure (time). In our previous studies of LDR irradiation in multiple tumor cells we show that all cell lines change in their rates of inactivation at circa 20 to 24 hours, so duration of exposure and dose-rate are both factors in

The data in figure 5 suggests that dose rates in the range of 0.25 to 0.49 Gy/hr increase tumor cell inactivation to rates that exceed that can be achieved by the alpha response induced by HDR irradiation. Thus, these patterns of inactivation show that glioblastoma cells while resistant to radiotherapeutic protocols that use multiple fractions below circa 3

Together with J.A. Williams, we have shown that combining irradiation delivered by an implanted radioactive seed with concomitant external beam fractionated radiotherapy produces significant increases in tumor response (Williams JA et al 1998). This study established the feasibility of combining brachytherapy and external beam radiotherapy to

**3.4 Radiosensitivity of xenograft tumors comprised of two R cells compared to a** 

The response of xenograft tumors that differ in their susceptibility to clonogenic inactivation in vitro also vary in their radiosensitivity to different radiotherapy protocols delivered in

**Dose Rate**

**0.086Gy/hr 0.123Gy/hr 0.25 Gy/hr 0.49 Gy/hr 0.49 Gy/hr alpha (HDR)**

**-0.35**

lower doses (alpha) and higher doses (omega).

achieving changes in clonogenic inactivation.

Gy, are more sensitive to protracted irradiation.

**radioresistant glioblastoma cell line** 

vivo.

achieve good responses in radioresistant glioblastoma cells.

**-0.3**

**-0.25**

**-0.2**

**-0.15**

**Log SF/Gy**

**-0.1**

**-0.05**

**0**

Fig. 4. Clonogenic survival patterns for two glioblastoma cell lines (U251, U87) and two colorectal tumor cell lines (HCT116 and DLD-1) irradiated with either HDR (50 Gy/hr) and LDR (0.25 Gy/hr). Data for each cell line are shown as two panels, the upper panel shows surviving fraction, lower panels show the rate of cell killing calculated as logs killed per Gy between sequential time points. The slopes of cell killing curves are represented by: α(HDR) which is measured from the slope of the line from 0.0 to surviving fraction at 2.0 Gy HDR; α (LDR) which is the slope of the line from 0.0 to 6.0 Gy LDR; the slope of the cell killing curve at doses greater than 4 Gy HDR determined by linear regression, ω (HDR); and the slope of cell killing at LDR doses greater than 6 Gy, ω (LDR). In the lower panels, the rate of cell killing by HDR is indicated by dotted, dashed lines.

These data show that both glioblastoma cell lines are inactivated by low dose-irradiation at approximately the same rates similar to rates for both colorectal tumor cells, showing a clear increase in rate that surpasses the levels of high dose-rate inactivation (omega response) of both cell types. This rate of LDR inactivation surpasses the rate of inactivation for the omega response for U-251 but the elevated level of U-87 cells for the omega response is not achieved.

These data suggest strongly that for radioresistant glioblastoma cells, the rate of inactivation by LDR irradiation can surpass the rate of inactivation for by multiple fractions that induce inactivation along the alpha response.

This in turn suggests using LDR radiotherapy for glioblastoma tumors and we synthesize our data to demonstrate the relative effect of different dose rates for induction of clonogenic inactivation. These data are shown in figure 5.

**-4 -3 -2 -1 0 1**

**-0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0**

achieved.

**ln(SF) / Gy**

**HDR LDR alphaD (HDR) omegaD (HDR) omegaD (LDR)**

**Log SF**

**0 5 10 15 20 25**

**DLD1**

**Dose(Gy)**

**0 12 24 36 48 60 72 84 96**

**ln SF/Gy alpha(HDR) omega(HDR)** **-0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0**

**-4 -3 -2 -1 0 1**

**Log SF**

**ln(SF) / Gy**

Fig. 4. Clonogenic survival patterns for two glioblastoma cell lines (U251, U87) and two colorectal tumor cell lines (HCT116 and DLD-1) irradiated with either HDR (50 Gy/hr) and LDR (0.25 Gy/hr). Data for each cell line are shown as two panels, the upper panel shows surviving fraction, lower panels show the rate of cell killing calculated as logs killed per Gy between sequential time points. The slopes of cell killing curves are represented by: α(HDR) which is measured from the slope of the line from 0.0 to surviving fraction at 2.0 Gy HDR; α (LDR) which is the slope of the line from 0.0 to 6.0 Gy LDR; the slope of the cell killing curve at doses greater than 4 Gy HDR determined by linear regression, ω (HDR); and the slope of cell killing at LDR doses greater than 6 Gy, ω (LDR). In the lower panels, the rate of cell

These data show that both glioblastoma cell lines are inactivated by low dose-irradiation at approximately the same rates similar to rates for both colorectal tumor cells, showing a clear increase in rate that surpasses the levels of high dose-rate inactivation (omega response) of both cell types. This rate of LDR inactivation surpasses the rate of inactivation for the omega response for U-251 but the elevated level of U-87 cells for the omega response is not

These data suggest strongly that for radioresistant glioblastoma cells, the rate of inactivation by LDR irradiation can surpass the rate of inactivation for by multiple fractions that induce

This in turn suggests using LDR radiotherapy for glioblastoma tumors and we synthesize our data to demonstrate the relative effect of different dose rates for induction of clonogenic

**0 12 24 36 48 60 72 84 96**

**Dose(Gy)**

**0 5 10 15 20 25**

**19S186**

**ln SF/Gy alpha(HDR) omega(HDR)**

**HDR LDR alphaD (HDR) omegaD (HDR) omegaD (LDR)**

**Time (hour)**

**Time (hour)**

killing by HDR is indicated by dotted, dashed lines.

inactivation along the alpha response.

inactivation. These data are shown in figure 5.

Fig. 5. Changes in the rate of cell killing expressed as Log10 of cells killed per Gy as a function of time of protracted irradiation in U251 cells irradiated with 4 different constant dose-rates and one that begins at 0.49 Gy/hr and decays with a half life of 2.7 days. The horizontal dotted and dashed lines represent the rate of cell killing for HDR irradiation at lower doses (alpha) and higher doses (omega).

These data show that increasing dose-rate increases the rate at which cells are inactivated until dose-rates reach approximately 0.12 Gy/hr when there is a relatively common response for higher dose-rates including our chosen LDR of 0.25 Gy/hr. The diagonal dashed lines in figure 5 are a general indication of the relative effects of dose-rate compared to the relative effects of duration of exposure (time). In our previous studies of LDR irradiation in multiple tumor cells we show that all cell lines change in their rates of inactivation at circa 20 to 24 hours, so duration of exposure and dose-rate are both factors in achieving changes in clonogenic inactivation.

The data in figure 5 suggests that dose rates in the range of 0.25 to 0.49 Gy/hr increase tumor cell inactivation to rates that exceed that can be achieved by the alpha response induced by HDR irradiation. Thus, these patterns of inactivation show that glioblastoma cells while resistant to radiotherapeutic protocols that use multiple fractions below circa 3 Gy, are more sensitive to protracted irradiation.

Together with J.A. Williams, we have shown that combining irradiation delivered by an implanted radioactive seed with concomitant external beam fractionated radiotherapy produces significant increases in tumor response (Williams JA et al 1998). This study established the feasibility of combining brachytherapy and external beam radiotherapy to achieve good responses in radioresistant glioblastoma cells.

#### **3.4 Radiosensitivity of xenograft tumors comprised of two R cells compared to a radioresistant glioblastoma cell line**

The response of xenograft tumors that differ in their susceptibility to clonogenic inactivation in vitro also vary in their radiosensitivity to different radiotherapy protocols delivered in vivo.

Radiobiology of Radioresistant Glioblastoma 17

**Comparison of** ρ **and** τ

**0123456**

VR

**U251 DLD1 19S186**

ρ **(in vivo radiosensitivity)**

Fig. 7. Relative in vitro radiosensitivity (τ) and in vivo radiosensitivity (ρ) for two R cells, DLD-1 and 19S186 and a radioresistant glioblastoma cells U-251. Each data point represents tumor response of each cell line to one of four protocols. Data point representing the two fractionated protocols, 8 x 2 and 2 x 5, are connected by a dashed lines and these responses are very similar for U-251 cells but are markedly different for the two R cell lines. The response for each cell line to a single dose of 7.5 Gy is connected by an arrow to response of the same tumor to a single dose of 15 Gy. These increased responses are significant beyond

These data show important differences between the response of glioblastoma cell line and the two R cell lines. First, the contribution of in vivo radiosensitivity represented by ρ to responses of the glioblastoma line is remarkably greater than the contribution of this sensitivity in R cells. In contradistinction, the values for the in vitro component of tumor radiosensitivity in vitro, τ, are diminished, this diminution similar to differences predicted by in vitro clonogenic inactivation. The effects shown in this figure are large up to a factor of 50 to 100 in doses to induce equivalent regression or in doses needed to induce the same

Our studies show that cell lines designated as "glioblastoma" in the literature are diverse in

**4.1.1 Some glioblastoma cell lines express a low rate of inactivation over the alpha\*** 

In our studies these reduced rates underlay the observation that such cell lines are refractory to doses over the alpha response. Specifically we hypothesized that these cell lines, express

**4.1 Glioblastoma cells express a diverse radiosensitivity phenotype** 

**0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4**

R

the differences observed in fractionated protocols.

τ

levels of regression.

**4. Conclusions** 

**response** 

their radiosensitivity phenotypes.

**(in vitro radiosensitivity)**

We performed a large set of experiments that compared the response of eight different cells that vary in their in vitro radiosensitivity to different radiotherapy protocols in vivo (Williams et al 2010). These studies showed a strong correlation between the total cells killed in vitro with tumor response but also showed a new in vivo effect that resulted from an in vivo interaction between tumor cell genotype and tumor microenvironment. Response of xenograft tumors comprised of two R cell lines, DLD-1 and 19S186 cells are compared to the very resistant VR line U-251 in figure 6.

Fig. 6. Growth of cohorts of xenograft tumors comprised of DLD-1, 19S186 and U-251 glioblastoma cells to four radiation protocols: 8 x 2 Gy, 2 x 5Gy, 1 x 7.5 Gy and 1 x 15 Gy. The ordinate in these figures represent the median log V/Vo. Tumors that were irradiated at approximately 0.2 gm and their volumes measured with time. Tumor size is expressed as median tumor volume as a function of days after irradiation.

The data in figure 6 show an increase in tumor radiosensitivity compared to their in vitro radiosensitivity for glioblastoma cells.

In detailed analysis of 40 experiments similar to those shown in figure 6, we showed tumor response could be resolved into two independent sensitivity factors: **τ** and **ρ** (23 ). The factor τ is related to total cells inactivated in vitro by each protocol and is dependent on genotype, fraction-size ant total dose. The factor ρ is dependent on genotype, fraction-size and total dose, but is independent of τ. We showed that for each protocol and genotype tumor response was dependent on the product of τ and ρ.

The relationship between τ and ρ is a useful comparison for radiosensitivity of xenograft tumors induced in tumor that vary in genotype and treated with different protocols. These coefficients are shown in figure 7 for DLD-1, 19S186 and U-251 cells.

We performed a large set of experiments that compared the response of eight different cells that vary in their in vitro radiosensitivity to different radiotherapy protocols in vivo (Williams et al 2010). These studies showed a strong correlation between the total cells killed in vitro with tumor response but also showed a new in vivo effect that resulted from an in vivo interaction between tumor cell genotype and tumor microenvironment. Response of xenograft tumors comprised of two R cell lines, DLD-1 and 19S186 cells are compared to the

> **-2 -1.5 -1 -0.5 0 0.5 1 1.5**

**Median**

Fig. 6. Growth of cohorts of xenograft tumors comprised of DLD-1, 19S186 and U-251 glioblastoma cells to four radiation protocols: 8 x 2 Gy, 2 x 5Gy, 1 x 7.5 Gy and 1 x 15 Gy. The ordinate in these figures represent the median log V/Vo. Tumors that were irradiated at approximately 0.2 gm and their volumes measured with time. Tumor size is expressed as

The data in figure 6 show an increase in tumor radiosensitivity compared to their in vitro

In detailed analysis of 40 experiments similar to those shown in figure 6, we showed tumor response could be resolved into two independent sensitivity factors: **τ** and **ρ** (23 ). The factor τ is related to total cells inactivated in vitro by each protocol and is dependent on genotype, fraction-size ant total dose. The factor ρ is dependent on genotype, fraction-size and total dose, but is independent of τ. We showed that for each protocol and genotype tumor

The relationship between τ and ρ is a useful comparison for radiosensitivity of xenograft tumors induced in tumor that vary in genotype and treated with different protocols. These

**Tumor Volum**

**0 20 40 60**

**19S186**

**Days**

**Control 7.5Gy 2x5Gy 15Gy 8x2Gy 4 x Vo**

**-2 -1.5 -1 -0.5 0 0.5 1 1.5 -40 60**

very resistant VR line U-251 in figure 6.

**DLD-1**

**0 20 40 60**

radiosensitivity for glioblastoma cells.

**U251**

**0 20 40 60**

**Days**

**Days**

response was dependent on the product of τ and ρ.

coefficients are shown in figure 7 for DLD-1, 19S186 and U-251 cells.

median tumor volume as a function of days after irradiation.

**-2 -1.5 -1 -0.5 0 0.5 1 1.5**

**-2 -1.5 -1 -0.5 0 0.5 1 1.5**

**Median Tumor Volum**

**Median Tumor Volum**

Fig. 7. Relative in vitro radiosensitivity (τ) and in vivo radiosensitivity (ρ) for two R cells, DLD-1 and 19S186 and a radioresistant glioblastoma cells U-251. Each data point represents tumor response of each cell line to one of four protocols. Data point representing the two fractionated protocols, 8 x 2 and 2 x 5, are connected by a dashed lines and these responses are very similar for U-251 cells but are markedly different for the two R cell lines. The response for each cell line to a single dose of 7.5 Gy is connected by an arrow to response of the same tumor to a single dose of 15 Gy. These increased responses are significant beyond the differences observed in fractionated protocols.

These data show important differences between the response of glioblastoma cell line and the two R cell lines. First, the contribution of in vivo radiosensitivity represented by ρ to responses of the glioblastoma line is remarkably greater than the contribution of this sensitivity in R cells. In contradistinction, the values for the in vitro component of tumor radiosensitivity in vitro, τ, are diminished, this diminution similar to differences predicted by in vitro clonogenic inactivation. The effects shown in this figure are large up to a factor of 50 to 100 in doses to induce equivalent regression or in doses needed to induce the same levels of regression.
