Impact of variation in the α/β of cervical cancer on predicted clinical outcomes.
To improve the effectiveness and efficiency in radiation therapy, various treatment modalities and fractionation schemes have been introduced and combined for cancer treatment. Biological effective dose (BED) and equivalent dose in 2 Gy fraction (EQD2) are used to evaluate and compare different modalities and fractionations and also used to determine dose prescriptions for new radiation schemes. BED and EQD2 are functions of α/β and the accuracy of α/β value is essential. A single α/β value of 10 Gy has been used for cervical cancer in clinical practice. However, our previous study first found that cervical cancer has a broad range of α/β values across in vitro studies that follow a right-skewed log-normal distribution. If patient populations follow such a distribution, it may have potential impact on radiation therapy for cervical cancer.
To investigate the impact of variation in the α/β of cervical cancer on the expected EQD2 associated with clinical outcome for cervix cancer patients treated with radiation therapy and how that variance could influence the determination of alternate fractionation schemes.
A right-skewed log-normal distribution of experimentally derived α/β values was applied to a reference tumor control probability (TCP) curve generated from cervical cancer patients treated with radiation, and a population of patients following that distribution were simulated using Monte Carlo sampling. An alternate equation for equivalent dose in 2 Gy fractions (EQD2) was derived that considered variance in α/β and was used to generate new values and associated TCP curves that could be plotted on a common EQD2 axis. Convolution curves of TCP and normal tissue complication probability (NTCP) were generated to determine the potential shift in optimal dose and probability of risk-free local control (RFLC). Theoretical treatment failure rates were generated to evaluate changes in rates of treatment outcomes.
Variation in α/β obtained from published experimental results produced potential losses in TCP of up to 24% in the range of clinical interest. RFLC curves predicted an optimal treatment dose of 95 Gy EQD2 when applying our most probable α/β of 4.25 Gy, 10 Gy higher than that predicted by the reference curve. The α/β distribution saw a decrease in RFLC of 17%. To achieve a TCP of 90%, possible HDR fractionation schemes ranged from 56 Gy in 14 fractions to 32 Gy in 2 fractions, with the associated increase in normal tissue dose ranging from 11 to 16 Gy EQD2.
The distribution of cervical cancer α/β values derived from experimental results produced significant changes in tumor control when applied to a reference TCP curve. TCP decreased with both the average and most probable α/β values. It is suggested that variance and heterogeneity in α/β be more explicitly incorporated to account for those patients that do deviate from the assumed constant value, especially in the case of evaluating different radiation schemes.
To investigate the impact of variation in the α/β of cervical cancer on the expected EQD2 associated with clinical outcome for cervix cancer patients treated with radiation therapy and how that variance could influence the determination of alternate fractionation schemes.
A right-skewed log-normal distribution of experimentally derived α/β values was applied to a reference tumor control probability (TCP) curve generated from cervical cancer patients treated with radiation, and a population of patients following that distribution were simulated using Monte Carlo sampling. An alternate equation for equivalent dose in 2 Gy fractions (EQD2) was derived that considered variance in α/β and was used to generate new values and associated TCP curves that could be plotted on a common EQD2 axis. Convolution curves of TCP and normal tissue complication probability (NTCP) were generated to determine the potential shift in optimal dose and probability of risk-free local control (RFLC). Theoretical treatment failure rates were generated to evaluate changes in rates of treatment outcomes.
Variation in α/β obtained from published experimental results produced potential losses in TCP of up to 24% in the range of clinical interest. RFLC curves predicted an optimal treatment dose of 95 Gy EQD2 when applying our most probable α/β of 4.25 Gy, 10 Gy higher than that predicted by the reference curve. The α/β distribution saw a decrease in RFLC of 17%. To achieve a TCP of 90%, possible HDR fractionation schemes ranged from 56 Gy in 14 fractions to 32 Gy in 2 fractions, with the associated increase in normal tissue dose ranging from 11 to 16 Gy EQD2.
The distribution of cervical cancer α/β values derived from experimental results produced significant changes in tumor control when applied to a reference TCP curve. TCP decreased with both the average and most probable α/β values. It is suggested that variance and heterogeneity in α/β be more explicitly incorporated to account for those patients that do deviate from the assumed constant value, especially in the case of evaluating different radiation schemes.