Amy E. Pomeroy
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Amy E. Pomeroy has a total of 25 co-authors. Here are the top co-authors by number of shared publications:
- Adam Christopher Palmer
The University of North Carolina at Chapel Hill Department of Pharmacology, Computational Medicine Program, UNC Lineberger Comprehensive Cancer Center Chapel Hill, North Carolina
88 shared publications - Sarah C. Patterson
Department of Pharmacology, Computational Medicine Program, UNC Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill
37 shared publications - Ash A. Alizadeh
Division of Oncology, Department of Medicine, Stanford University School of Medicine, Stanford, USA
9 shared publications - Brian J. Sworder
Division of Hematology Oncology, Department of Medicine, University of California, Irvine School of Medicine, Irvine, CA
9 shared publications - Deborah Plana
Laboratory of Systems Pharmacology and the Department of Systems Biology, Harvard Medical School, Boston, Massachusetts
9 shared publications - Yanguang Cao
Division of Pharmacotherapy and Experimental Therapeutics, UNC Eshelman School of Pharmacy University of North Carolina at Chapel Hill Chapel Hill North Carolina USA
9 shared publications
Latest Publications
- 2 September 2026
- Published by American Association for Cancer Research (AACR) in Cancer Research
- Vol. 86 (17) , 4195-4196
Varying Cancer Drug Doses to Balance Tumor Control and Resistance Evolution
The effectiveness of any cancer therapy depends on its dose and its treatment schedule. A potential third dimension of regimen design is whether intentional fluctuations in administered dose might improve treatment outcomes. In this issue, West and colleagues demonstrate that dose fluctuations can alter both tumor drug response and the evolution of drug resistance, in ways related to the shapes of dose–response functions. In mouse xenograft models of ALK-fusion non–small cell lung cancer treated with the ALK inhibitor alectinib, steady dosing produced the best immediate tumor response, but fluctuations in dose were better at delaying the evolution of drug resistance. Treatment schedules that switched between even and uneven dosing navigated this tradeoff, achieving comparable tumor control with steady dosing while better preserving drug sensitivity. By showing how dose–response measurements can guide when doses should remain steady or fluctuate, this work expands how cancer treatment schedules might be optimized. See related article by West et al., p. 4434
- 2 September 2026
- Published by American Association for Cancer Research (AACR) in Cancer Research
- Vol. 86 (17) , 4195-4196
Varying Cancer Drug Doses to Balance Tumor Control and Resistance Evolution
The effectiveness of any cancer therapy depends on its dose and its treatment schedule. A potential third dimension of regimen design is whether intentional fluctuations in administered dose might improve treatment outcomes. In this issue, West and colleagues demonstrate that dose fluctuations can alter both tumor drug response and the evolution of drug resistance, in ways related to the shapes of dose–response functions. In mouse xenograft models of ALK-fusion non–small cell lung cancer treated with the ALK inhibitor alectinib, steady dosing produced the best immediate tumor response, but fluctuations in dose were better at delaying the evolution of drug resistance. Treatment schedules that switched between even and uneven dosing navigated this tradeoff, achieving comparable tumor control with steady dosing while better preserving drug sensitivity. By showing how dose–response measurements can guide when doses should remain steady or fluctuate, this work expands how cancer treatment schedules might be optimized. See related article by West et al., p. 4434
- 3 July 2026
- Published by American Association for Cancer Research (AACR) in Molecular Cancer Therapeutics
- p. OF1-OF14
Functional Genomics and Proteomics Identify Folate Carrier SLC19A1 as a Predictor of Pralatrexate Sensitivity in Diverse T-cell Lymphoma Models
Cancer therapies are typically effective in subsets of patients, reflecting the molecular diversity of cancers and motivating the need for predictive biomarkers of response. Biomarker-guided therapy is increasingly useful in oncology, yet biomarker discovery remains complicated by the large number of molecular features that make it difficult to distinguish causal determinants from spurious associations. To address this challenge, we combined functional genomic screening, proteomics, and drug sensitivity profiling to discover response biomarkers for a number of therapies used in the treatment of peripheral T-cell lymphomas (PTCL). First, we used genome-wide CRISPR–dCas9 interference screens in PTCL cells under drug treatment to identify a shortlist of genes whose knockdown directly increases or decreases drug sensitivity. Next, we profiled drug responses across a diverse panel of 30 PTCL cell cultures and, from the shortlist, identified genes whose protein abundance correlated with drug sensitivity. Genes detected by both approaches are causal determinants of drug response and correlates of drug response across the panel of cultures, making them promising candidates for predictive biomarkers. Basal expression of the reduced folate carrier SLC19A1was a strong predictor of pralatrexate sensitivity, consistent with its role as the primary transporter for pralatrexate uptake. Simulated clinical trials predicted that biomarker-guided patient selection could improve the power to detect significant benefit of adding pralatrexate to first-line chemotherapy in PTCL. These findings illustrate how functional genetic screens can augment correlative studies to identify candidate biomarkers of drug response and suggest the potential for precise use of pralatrexate for PTCL.
- 3 July 2026
- Published by American Association for Cancer Research (AACR) in Molecular Cancer Therapeutics
- p. OF1-OF14
Functional Genomics and Proteomics Identify Folate Carrier SLC19A1 as a Predictor of Pralatrexate Sensitivity in Diverse T-cell Lymphoma Models
Cancer therapies are typically effective in subsets of patients, reflecting the molecular diversity of cancers and motivating the need for predictive biomarkers of response. Biomarker-guided therapy is increasingly useful in oncology, yet biomarker discovery remains complicated by the large number of molecular features that make it difficult to distinguish causal determinants from spurious associations. To address this challenge, we combined functional genomic screening, proteomics, and drug sensitivity profiling to discover response biomarkers for a number of therapies used in the treatment of peripheral T-cell lymphomas (PTCL). First, we used genome-wide CRISPR–dCas9 interference screens in PTCL cells under drug treatment to identify a shortlist of genes whose knockdown directly increases or decreases drug sensitivity. Next, we profiled drug responses across a diverse panel of 30 PTCL cell cultures and, from the shortlist, identified genes whose protein abundance correlated with drug sensitivity. Genes detected by both approaches are causal determinants of drug response and correlates of drug response across the panel of cultures, making them promising candidates for predictive biomarkers. Basal expression of the reduced folate carrier SLC19A1was a strong predictor of pralatrexate sensitivity, consistent with its role as the primary transporter for pralatrexate uptake. Simulated clinical trials predicted that biomarker-guided patient selection could improve the power to detect significant benefit of adding pralatrexate to first-line chemotherapy in PTCL. These findings illustrate how functional genetic screens can augment correlative studies to identify candidate biomarkers of drug response and suggest the potential for precise use of pralatrexate for PTCL.
- Open Access
- 17 June 2026
- Published by American Association for Cancer Research (AACR)
Supplementary Figure 2 from Ultrasensitive Response Explains the Benefit of Combination Chemotherapy Despite Drug Antagonism
Supplemental Figure 2. Isobologram analysis shows antagonism between CH and HO in the CHOP regimen for Peripheral T-Cell Lymphoma. A) Left: shapes of equal-effect contour lines (isoboles) reveal non-additive drug interactions (Palmer et al.)9. Right: Cell viability is measured across an 11x11 concentration gradient to graph isobolograms. B) CH, HO, and CO drug interactions in four Peripheral T-Cell Lymphoma (PTCL) cell lines: KHYG-1, Ki-JK, MTA, SU-DHL-1 (n=4 replicates per plot). White contour lines highlight 50%, 20%, and 5% relative viability. Green dots mark a clinically relevant concentration, Csustained (the average plasma concentration in humans 6 hours after administration12). C) CH and HO pairwise drug interactions in three additional PTCL cell lines: SMZ-1, MOT-N1, DL-40 (n=4 replicates per plot).
- 17 June 2026
- Published by American Association for Cancer Research (AACR)
Open AccessSupplementary Figure 2 from Ultrasensitive Response Explains the Benefit of Combination Chemotherapy Despite Drug Antagonism
Supplemental Figure 2. Isobologram analysis shows antagonism between CH and HO in the CHOP regimen for Peripheral T-Cell Lymphoma. A) Left: shapes of equal-effect contour lines (isoboles) reveal non-additive drug interactions (Palmer et al.)9. Right: Cell viability is measured across an 11x11 concentration gradient to graph isobolograms. B) CH, HO, and CO drug interactions in four Peripheral T-Cell Lymphoma (PTCL) cell lines: KHYG-1, Ki-JK, MTA, SU-DHL-1 (n=4 replicates per plot). White contour lines highlight 50%, 20%, and 5% relative viability. Green dots mark a clinically relevant concentration, Csustained (the average plasma concentration in humans 6 hours after administration12). C) CH and HO pairwise drug interactions in three additional PTCL cell lines: SMZ-1, MOT-N1, DL-40 (n=4 replicates per plot).
- Open Access
- 17 June 2026
- Published by American Association for Cancer Research (AACR)
Supplementary Figure 4 from Ultrasensitive Response Explains the Benefit of Combination Chemotherapy Despite Drug Antagonism
Supplemental Figure 4. Drug interactions in CHOP using MuSyC method to distinguish potency and efficacy shifts. A) The MuSyC method analyzes drug combination effects to distinguish between shifts in potency and shifts in efficacy. Shifts in potency from the Bliss model (orange) to observed data (blue) are represented by parameter A, where positive values indicate superior potency and negative values indicate inferior potency. Changes in efficacy are represented by parameter B, where positive values indicate greater inhibition, and negative values indicate less inhibition. B) Example of method for a drug combination with lower efficacy than predicted by the Bliss model. The interaction model (red) is fit from the Bliss model to the experimental data across the length of the curve. The quadrant plot shows the A and B parameters that confer the best fit determined by sum of squares errors. Stronger fits are colored darker blue and the best fit is marked in red. Data shows drug combination CHOP in SU-DHL-1 T-Cell lymphoma cell line, n=8. C) Example of method for a drug combination with higher efficacy than Bliss model predicts. Data shows drug combination leuocorovin-fluorouracil-irinotecan in Lovo Colorectal cancer cell line, n=8. D) Shifts in potency and efficacy for all 11 combinations of agents in CHOP, for each of four Peripheral T-Cell Lymphoma cell lines: KHYG-1, SU-DHL-1, Ki-JK, MTA. Yellow region marks a two-fold difference in potency or efficacy compared with the ‘no interaction’ null hypothesis. Error bars are 95% confidence intervals, n=8.
- 17 June 2026
- Published by American Association for Cancer Research (AACR)
Open AccessSupplementary Figure 4 from Ultrasensitive Response Explains the Benefit of Combination Chemotherapy Despite Drug Antagonism
Supplemental Figure 4. Drug interactions in CHOP using MuSyC method to distinguish potency and efficacy shifts. A) The MuSyC method analyzes drug combination effects to distinguish between shifts in potency and shifts in efficacy. Shifts in potency from the Bliss model (orange) to observed data (blue) are represented by parameter A, where positive values indicate superior potency and negative values indicate inferior potency. Changes in efficacy are represented by parameter B, where positive values indicate greater inhibition, and negative values indicate less inhibition. B) Example of method for a drug combination with lower efficacy than predicted by the Bliss model. The interaction model (red) is fit from the Bliss model to the experimental data across the length of the curve. The quadrant plot shows the A and B parameters that confer the best fit determined by sum of squares errors. Stronger fits are colored darker blue and the best fit is marked in red. Data shows drug combination CHOP in SU-DHL-1 T-Cell lymphoma cell line, n=8. C) Example of method for a drug combination with higher efficacy than Bliss model predicts. Data shows drug combination leuocorovin-fluorouracil-irinotecan in Lovo Colorectal cancer cell line, n=8. D) Shifts in potency and efficacy for all 11 combinations of agents in CHOP, for each of four Peripheral T-Cell Lymphoma cell lines: KHYG-1, SU-DHL-1, Ki-JK, MTA. Yellow region marks a two-fold difference in potency or efficacy compared with the ‘no interaction’ null hypothesis. Error bars are 95% confidence intervals, n=8.
- Open Access
- 17 June 2026
- Published by American Association for Cancer Research (AACR)
Supplementary Figure 10 from Ultrasensitive Response Explains the Benefit of Combination Chemotherapy Despite Drug Antagonism
Supplemental Figure 10. Combinations of drugs with varied mechanisms applied to the linear-quadratic model. We present here a preliminary investigation of how the linear-quadratic model (LQ) could have different implications if multiple drugs exert more distinct mechanisms of killing. Different drugs may exhibit different parameter values for α and β in the LQ model, but a further complexity is whether the quadratic terms are shared or distinct, which is compared here. (A) In this article, we have modelled cyclophosphamide and doxorubicin as both causing DNA damage and jointly adding to the accumulation of damage that is represented by the ‘quadratic’ term in the LQ model. When this applies to a hypothetical combination of drugs A+B, their combined dose can reach the ultrasensitive region of response more readily. Mathematically, the sum of drugs’ effects is squared. Without formally introducing drug-drug interaction, this equation results in an isobologram plot showing drugs A+B to be additive. (B) If drugs with highly distinct mechanisms of action are combined, their forms of damage may not jointly add up in the quadratic term of the LQ model. It may instead be justified to introduce distinct quadratic terms for each drug. Interestingly, even without formally introducing drug-drug interaction, such an equation leads to an isobologram plot that exhibits antagonistic (or sub-additive) interaction, even though the effect of the drug combination has been modelled as precisely additive. This arises because a combination of two half-doses is inferior to a full-dose of either single drug, because neither single drug dose extends as far into the ‘ultrasensitive’ region of response.
- 17 June 2026
- Published by American Association for Cancer Research (AACR)
Open AccessSupplementary Figure 10 from Ultrasensitive Response Explains the Benefit of Combination Chemotherapy Despite Drug Antagonism
Supplemental Figure 10. Combinations of drugs with varied mechanisms applied to the linear-quadratic model. We present here a preliminary investigation of how the linear-quadratic model (LQ) could have different implications if multiple drugs exert more distinct mechanisms of killing. Different drugs may exhibit different parameter values for α and β in the LQ model, but a further complexity is whether the quadratic terms are shared or distinct, which is compared here. (A) In this article, we have modelled cyclophosphamide and doxorubicin as both causing DNA damage and jointly adding to the accumulation of damage that is represented by the ‘quadratic’ term in the LQ model. When this applies to a hypothetical combination of drugs A+B, their combined dose can reach the ultrasensitive region of response more readily. Mathematically, the sum of drugs’ effects is squared. Without formally introducing drug-drug interaction, this equation results in an isobologram plot showing drugs A+B to be additive. (B) If drugs with highly distinct mechanisms of action are combined, their forms of damage may not jointly add up in the quadratic term of the LQ model. It may instead be justified to introduce distinct quadratic terms for each drug. Interestingly, even without formally introducing drug-drug interaction, such an equation leads to an isobologram plot that exhibits antagonistic (or sub-additive) interaction, even though the effect of the drug combination has been modelled as precisely additive. This arises because a combination of two half-doses is inferior to a full-dose of either single drug, because neither single drug dose extends as far into the ‘ultrasensitive’ region of response.
- Open Access
- 17 June 2026
- Published by American Association for Cancer Research (AACR)
Supplementary Figure 8 from Ultrasensitive Response Explains the Benefit of Combination Chemotherapy Despite Drug Antagonism
Supplemental Figure 8. Model parameters informed from experimental data. A) The model’s death rate was determined by treating MTA cells with lethal doses of 4-H-cyclophophamide and doxorubicin and observing that the cell population halves every 2.5 days, n=1 per condition. Cells were incubated with drugs for 3 days (represented by yellow bar) before being washed off and replaced with drug-free media. B) The model’s treated growth rate was determined by treating MTA cells with the same drug concentrations used in the concurrent vs sequential dosing experiments and observing that cells double every 2.5 days, n=1 per condition. Cells were incubated with drugs for 3 days (represented by yellow bar) before being washed off and replaced with drug-free media. C) At each simulated drug treatment (marked with yellow triangles), the model calculates the fraction of cells that will survive and the fraction that will die. Each respective cell population is modelled as growing or dying at the experimentally determined rates. Total live cell count is calculated by adding the number of growing cells with the number of dying cells over time. D) 4-H-cyclophosphamide yields about 20% viability and doxorubicin yields about 40% viability at 13% Csustained, the concentration of drug that was used in the concurrent vs sequential administration experiments (2 μM and 20 nM respectively), n=8 per condition. E) To convert experimental concentrations to dose values for the model, the corresponding dose values at 20% and 40% viability were estimated to be 1.8 for doxorubicin and 2.6 for 4-H-cyclophosphamide. F) The CH isobologram created by the model matches the experimentally determined isobologram for MTA cells in Supplemental Figure 2b when the drug interaction term is set to a sub-additive value of i = -0.8. G) Model outputs of tumor cell populations over time were fit to the experimental data to fit the level of ultrasensitivity (β = 0.4) and the level of heterogeneity in drug sensitivity (σ = 0.42).
- 17 June 2026
- Published by American Association for Cancer Research (AACR)
Open AccessSupplementary Figure 8 from Ultrasensitive Response Explains the Benefit of Combination Chemotherapy Despite Drug Antagonism
Supplemental Figure 8. Model parameters informed from experimental data. A) The model’s death rate was determined by treating MTA cells with lethal doses of 4-H-cyclophophamide and doxorubicin and observing that the cell population halves every 2.5 days, n=1 per condition. Cells were incubated with drugs for 3 days (represented by yellow bar) before being washed off and replaced with drug-free media. B) The model’s treated growth rate was determined by treating MTA cells with the same drug concentrations used in the concurrent vs sequential dosing experiments and observing that cells double every 2.5 days, n=1 per condition. Cells were incubated with drugs for 3 days (represented by yellow bar) before being washed off and replaced with drug-free media. C) At each simulated drug treatment (marked with yellow triangles), the model calculates the fraction of cells that will survive and the fraction that will die. Each respective cell population is modelled as growing or dying at the experimentally determined rates. Total live cell count is calculated by adding the number of growing cells with the number of dying cells over time. D) 4-H-cyclophosphamide yields about 20% viability and doxorubicin yields about 40% viability at 13% Csustained, the concentration of drug that was used in the concurrent vs sequential administration experiments (2 μM and 20 nM respectively), n=8 per condition. E) To convert experimental concentrations to dose values for the model, the corresponding dose values at 20% and 40% viability were estimated to be 1.8 for doxorubicin and 2.6 for 4-H-cyclophosphamide. F) The CH isobologram created by the model matches the experimentally determined isobologram for MTA cells in Supplemental Figure 2b when the drug interaction term is set to a sub-additive value of i = -0.8. G) Model outputs of tumor cell populations over time were fit to the experimental data to fit the level of ultrasensitivity (β = 0.4) and the level of heterogeneity in drug sensitivity (σ = 0.42).
- Open Access
- 17 June 2026
- Published by American Association for Cancer Research (AACR)
Supplementary Figure S6 from A Model of Intratumor and Interpatient Heterogeneity Explains Clinical Trials of Curative Combination Therapy for Lymphoma
Treatment with RCHOP decreases patients’ sensitivities to RCHOP and novel agents
- 17 June 2026
- Published by American Association for Cancer Research (AACR)
Open AccessSupplementary Figure S6 from A Model of Intratumor and Interpatient Heterogeneity Explains Clinical Trials of Curative Combination Therapy for Lymphoma
Treatment with RCHOP decreases patients’ sensitivities to RCHOP and novel agents
- Open Access
- 17 June 2026
- Published by American Association for Cancer Research (AACR)
Supplementary Figure 9 from Ultrasensitive Response Explains the Benefit of Combination Chemotherapy Despite Drug Antagonism
Supplemental Figure 9. Model including a distribution of growth rates. The model presented in Figures 4 and 5 assigns cells to one of two outcomes after drug treatment: growing at a consistent rate, or dying at a consistent rate. In reality cells may experience more varied responses to treatment including various growth and death rates as well as cytostatic responses. To confirm that our experimental observations and conclusions about concurrent versus sequential therapy are robust to the possibility of graded drug responses, we here present a more complex model in which increasing drug doses causes, at first, a gradual decline in growth rate, then growth arrest, and then an increasing death rate. A) Each cell is assigned a unique growth rate (GR) where GR = 1 represents living cells with a doubling time of 2.5 days, GR = 0 represents cytostatic cells, and GR = -1 represents dying cells with a half-life of 2.5 days. GR is modelled as increasing linearly with drug dose, d, such that Growth Rate(d)=1-(1+Maximum Killing)*d where Maximum Killing = -1. Maximum rate of killing could in principle be faster, but experimental measurements shows that the rate of death after high doses of doxorubicin and 4-H-cyclophosphamide reaches a plateau that is similar to the uninhibited rate of growth (but for being a negative rate instead of a positive rate), which corresponds to GR = -1 (Supplemental Figure 8A). Here d represents effective combined dose accounting for drug interactions, ultrasensitivity, and cell-to-cell heterogeneity in sensitivity. Specifically, as a function of single-drug doses a and b, drug interaction i, LQ model parameters α and β, and relative drug sensitivity x, the combined effective drug dose is d = 10^x (alpha (a + b + (i a b) / (a+b)) + beta (a + b + (i a b) / (a+b))^2 ) Parameters defining ultrasensitivity (α and β), drug-drug interaction (i) and cellular heterogeneity (x) are unchanged from Figures 4 and 5. Recall that in a population of tumor cells, x is normally distributed around 0; therefore 10^x is log-normally distributed around 1 and describes cell-to-cell heterogeneity in drug sensitivity, rendering the combination therapy more or less potent against individual cells. The outcome of this model is that given doses of two drugs a and b, cancer cells in a population experience a variety of GR values spanning from 1 to -1, meaning that some cells are partially growth inhibited, some are arrested, and some are dying at various rates. After an initial drug treatment, an individual cell’s GR value (rate of growth or death) is constant until day 9, unless a sequential treatment on day 6 exerts a stronger effect that results in faster death or more severe growth inhibition. That is, the second dose can intensify response but not weaken it. B) This model with more diverse cellular responses to treatment also qualitatively reproduces the experimentally observed superiority of concurrent treatment over sequential treatment, when it includes ultrasensitivity and cellular heterogeneity. Yellow triangles mark simulated treatment times (concurrent: C and H both on day 0; sequential: C on day 0 and H on day 6).
- 17 June 2026
- Published by American Association for Cancer Research (AACR)
Open AccessSupplementary Figure 9 from Ultrasensitive Response Explains the Benefit of Combination Chemotherapy Despite Drug Antagonism
Supplemental Figure 9. Model including a distribution of growth rates. The model presented in Figures 4 and 5 assigns cells to one of two outcomes after drug treatment: growing at a consistent rate, or dying at a consistent rate. In reality cells may experience more varied responses to treatment including various growth and death rates as well as cytostatic responses. To confirm that our experimental observations and conclusions about concurrent versus sequential therapy are robust to the possibility of graded drug responses, we here present a more complex model in which increasing drug doses causes, at first, a gradual decline in growth rate, then growth arrest, and then an increasing death rate. A) Each cell is assigned a unique growth rate (GR) where GR = 1 represents living cells with a doubling time of 2.5 days, GR = 0 represents cytostatic cells, and GR = -1 represents dying cells with a half-life of 2.5 days. GR is modelled as increasing linearly with drug dose, d, such that Growth Rate(d)=1-(1+Maximum Killing)*d where Maximum Killing = -1. Maximum rate of killing could in principle be faster, but experimental measurements shows that the rate of death after high doses of doxorubicin and 4-H-cyclophosphamide reaches a plateau that is similar to the uninhibited rate of growth (but for being a negative rate instead of a positive rate), which corresponds to GR = -1 (Supplemental Figure 8A). Here d represents effective combined dose accounting for drug interactions, ultrasensitivity, and cell-to-cell heterogeneity in sensitivity. Specifically, as a function of single-drug doses a and b, drug interaction i, LQ model parameters α and β, and relative drug sensitivity x, the combined effective drug dose is d = 10^x (alpha (a + b + (i a b) / (a+b)) + beta (a + b + (i a b) / (a+b))^2 ) Parameters defining ultrasensitivity (α and β), drug-drug interaction (i) and cellular heterogeneity (x) are unchanged from Figures 4 and 5. Recall that in a population of tumor cells, x is normally distributed around 0; therefore 10^x is log-normally distributed around 1 and describes cell-to-cell heterogeneity in drug sensitivity, rendering the combination therapy more or less potent against individual cells. The outcome of this model is that given doses of two drugs a and b, cancer cells in a population experience a variety of GR values spanning from 1 to -1, meaning that some cells are partially growth inhibited, some are arrested, and some are dying at various rates. After an initial drug treatment, an individual cell’s GR value (rate of growth or death) is constant until day 9, unless a sequential treatment on day 6 exerts a stronger effect that results in faster death or more severe growth inhibition. That is, the second dose can intensify response but not weaken it. B) This model with more diverse cellular responses to treatment also qualitatively reproduces the experimentally observed superiority of concurrent treatment over sequential treatment, when it includes ultrasensitivity and cellular heterogeneity. Yellow triangles mark simulated treatment times (concurrent: C and H both on day 0; sequential: C on day 0 and H on day 6).
- Open Access
- 17 June 2026
- Published by American Association for Cancer Research (AACR)
Supplementary Figure 6 from Ultrasensitive Response Explains the Benefit of Combination Chemotherapy Despite Drug Antagonism
Supplemental Figure 6. Experimental population dose responses on log-kill scale. A) Many chemotherapies exhibit exponential dose responses in hematologic cancers (left), which appear linear when plotted on a logarithmic scale (right). The first 90% cytotoxic effect is shown in green (reaching 1 log-kill); the next 90% cytotoxic effect is shown in blue (reaching 2 log-kills); the next 90% cytotoxic effect is shown in red (reaching 3 log-kills). B) Log-kills produced by CHOP and its components at concentrations from 0 to 100% Csustained. n=8 for SU-DHL-1, KHYG-1, and Ki-JK; n=4 for SMZ-1, MOT-N1, and DL-40.
- 17 June 2026
- Published by American Association for Cancer Research (AACR)
Open AccessSupplementary Figure 6 from Ultrasensitive Response Explains the Benefit of Combination Chemotherapy Despite Drug Antagonism
Supplemental Figure 6. Experimental population dose responses on log-kill scale. A) Many chemotherapies exhibit exponential dose responses in hematologic cancers (left), which appear linear when plotted on a logarithmic scale (right). The first 90% cytotoxic effect is shown in green (reaching 1 log-kill); the next 90% cytotoxic effect is shown in blue (reaching 2 log-kills); the next 90% cytotoxic effect is shown in red (reaching 3 log-kills). B) Log-kills produced by CHOP and its components at concentrations from 0 to 100% Csustained. n=8 for SU-DHL-1, KHYG-1, and Ki-JK; n=4 for SMZ-1, MOT-N1, and DL-40.
- Open Access
- 17 June 2026
- Published by American Association for Cancer Research (AACR)
Supplementary Figure 3 from Ultrasensitive Response Explains the Benefit of Combination Chemotherapy Despite Drug Antagonism
Supplemental Figure 3. Analysis of drug pairs, triplets, and quadruplet CHOP confirms antagonistic interactions by the Bliss Independence model. A) Effect of CHOP combination measured at concentrations from 0 to 200% of Csustained, compared with Bliss Independence model computed from monotherapy effects, in four PTCL cell lines: KHYG-1, Ki-JK, MTA, SU-DHL-1. Blue shading represents 95% confidence interval (n=8 replicates). B) Relationship between observed cell viability, and predicted viability from Bliss Independence model, for all possible pairs, triplets, and quadruplets of drugs in CHOP across four PTCL cell lines (n=237 points across concentration range up to 500% Csustained; each point is an average of 8 measurements). All combinations containing antagonistic pairs CH or HO are colored in blue; all other combinations in red. Lines of best fit show that, as a group, combinations containing the drug pairs CH or HO (blue line) are antagonistic compared with the Bliss independence model (black line) (P = 6×10-34, Student’s t-test, n=117), and also more antagonistic than all combinations without CH or HO (red line) (P = 3×10-14, Student’s t-test, n=237). Overall, all CHOP measurements (blue and red combined) are antagonistic compared with the Bliss independence model (black line), with an average slope of 0.79 (P = 7×10-35, Student’s t-test, n=237). C) Histogram depicting the difference in observed and expected log-kills (-log10(relative viability) for all interactions involving CH or HO (blue), and all remaining interactions (red)).
- 17 June 2026
- Published by American Association for Cancer Research (AACR)
Open AccessSupplementary Figure 3 from Ultrasensitive Response Explains the Benefit of Combination Chemotherapy Despite Drug Antagonism
Supplemental Figure 3. Analysis of drug pairs, triplets, and quadruplet CHOP confirms antagonistic interactions by the Bliss Independence model. A) Effect of CHOP combination measured at concentrations from 0 to 200% of Csustained, compared with Bliss Independence model computed from monotherapy effects, in four PTCL cell lines: KHYG-1, Ki-JK, MTA, SU-DHL-1. Blue shading represents 95% confidence interval (n=8 replicates). B) Relationship between observed cell viability, and predicted viability from Bliss Independence model, for all possible pairs, triplets, and quadruplets of drugs in CHOP across four PTCL cell lines (n=237 points across concentration range up to 500% Csustained; each point is an average of 8 measurements). All combinations containing antagonistic pairs CH or HO are colored in blue; all other combinations in red. Lines of best fit show that, as a group, combinations containing the drug pairs CH or HO (blue line) are antagonistic compared with the Bliss independence model (black line) (P = 6×10-34, Student’s t-test, n=117), and also more antagonistic than all combinations without CH or HO (red line) (P = 3×10-14, Student’s t-test, n=237). Overall, all CHOP measurements (blue and red combined) are antagonistic compared with the Bliss independence model (black line), with an average slope of 0.79 (P = 7×10-35, Student’s t-test, n=237). C) Histogram depicting the difference in observed and expected log-kills (-log10(relative viability) for all interactions involving CH or HO (blue), and all remaining interactions (red)).