Robustness of Clonogenic Assays as a Biomarker for Cancer Cell Radiosensitivity

Abstract

Photon radiation therapy is a major curative treatment for cancer. However, the lack of robust predictive biomarkers for radiosensitivity precludes personalized radiation therapy. Clonogenic assays are the gold standard method for measuring the radiosensitivity of cancer cells. Although a large number of publications describe the use of clonogenic assays to measure cancer cell radiosensitivity, the robustness of results from different studies is unclear. To address this, we conducted a comprehensive detailed literature search of 256 common cancer cell lines and identified the eight cell lines most-frequently examined for photon sensitivity using clonogenic assays. Survival endpoints and experimental parameters from all 620 relevant experiments were compiled and analyzed. We found that the coefficients of variation for SF₂ (surviving fraction after 2 Gy irradiation) and for D₁₀ (dose that yields a surviving fraction of 10%) were below 30% for all cell lines, indicating that SF₂ and D₁₀ have acceptable inter-assay precision. These data support further analysis of published data on clonogenic assays using SF₂ and D₁₀ as survival endpoints, which facilitates robust identification of biological profiles representative of cancer cell sensitivity to photons.

Keywords: cancer; clonogenic assays; precision medicine; radiation therapy; radiosensitivity.

Figure 1

Frequency histograms of the LQ parameters α (a), β (b), and α/β (c), and a scatter plot showing correlation between α and β (d). R value and p value assessed by Spearman’s rank correlation test are shown.

Figure 2

Summary of clonogenic assay data acquired from the literature. (a) Timing of cell seeding. (b) Radiation type. (c) Dose rate. (d) SF₂. (e) SF₄. (f) SF₆. (g) SF₈. (h) D₁₀. (i) D₅₀. (j) $\overline{D}$. IR, irradiation.

Figure 3

CV values of clonogenic survival endpoints for the cancer cell lines. The dashed line indicates a CV of 30%.

Figure 4

Correlation between CV values and corresponding clonogenic survival endpoints. R values and p values (assessed by Spearman’s rank correlation test) are shown.

Figure 5

Univariate and multivariate analyses of the influence of experimental setting on clonogenic survival. For univariate analysis, differences in SF₂, SF₄, SF₆, SF₈, D₁₀, D₅₀, or the mean inactivation dose (MID) between different timings of cell seeding (i.e., before irradiation vs. after irradiation) or between different radiation types (i.e., X-rays vs. γ-rays) were examined using the Mann–Whitney U test. The correlation between dose rate and SF₂, SF₄, SF₆, SF₈, D₁₀, D₅₀, or $\overline{D}$ was examined using Spearman’s rank correlation test. For multivariate analysis, the effect of timing of cell seeding, radiation type, and dose rate on SF₂, SF₄, SF₆, SF₈, D₁₀, D₅₀, or $\overline{D}$ was examined by multiple linear regression. Original panels show statistical significance (p value < 0.05) after initial analyses. BH and Bonferroni panels show statistical significance after Benjamini–Hockberg and Bonferroni correction, respectively. Black panels, statistical test not applicable due to insufficient sample size.

Figure 6

Scheme for the literature search.