Prediction of ultra-high-order antibiotic combinations based on pairwise interactions

Katzir, Itay; Çokol, Murat; Aldridge, Bree B.; Alon, Uri

doi:10.1371/journal.pcbi.1006774

Cited by 59 publications

(74 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In retrospect this is not surprising, as we used a different genotype of E. coli than was used by Yeh et al, and minimal rather than rich growth medium. Additionally, we used a yield-based checkerboard assay, while they used the growth rate-based dose-response curve measurement method ( 12 ). We elected to do a yield-based method because it allowed us to more highly parallelize our experiments.…”

Section: Discussionmentioning

confidence: 99%

“…However, clinical trials of combination therapy in the treatment of bacterial infections in patients have been limited. Choosing the correct drug combination is difficult ( 12 , 13 ), and efficacy has been mixed ( 14 , 15 ). A greater understanding of the mechanisms driving effective combination therapy are therefore required for successful clinical implementation.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Weakest-Link Dynamics Predict Apparent Antibiotic Interactions in a Model Cross-Feeding Community

Adamowicz

Harcombe

2020

Antimicrob Agents Chemother

View full text Add to dashboard Cite

With the growing global threat of antimicrobial resistance, novel strategies are required for combatting resistant pathogens. Combination therapy, wherein multiple drugs are used to treat an infection, has proven highly successful in the treatment of cancer and HIV. However, this practice has proven challenging for the treatment of bacterial infections due to difficulties in selecting the correct combinations and dosages. An additional challenge in infection treatment is the polymicrobial nature of many infections, which may respond to antibiotics differently than a monoculture pathogen. This study tests whether patterns of antibiotic interactions (synergy, antagonism, or independence/additivity) in monoculture can be used to predict antibiotic interactions in an obligate cross-feeding co-culture. Using our previously described weakest link hypothesis, we hypothesized antibiotic interactions in co-culture based on the interactions we observed in monoculture. We then compared our predictions to observed antibiotic interactions in co-culture. We tested the interactions between ten previously identified antibiotic combinations using checkerboard assays. Although our antibiotic combinations interacted differently than predicted in our monocultures, our monoculture results were generally sufficient to predict co-culture patterns based solely on the weakest link hypothesis. These results suggest that combination therapy for cross-feeding multispecies infections may be successfully designed based on antibiotic interaction patterns for their component species.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Weakest-Link Dynamics Predict Apparent Antibiotic Interactions in a Model Cross-Feeding Community

Adamowicz

Harcombe

2020

Antimicrob Agents Chemother

View full text Add to dashboard Cite

show abstract

“…In previous in vitro studies, we harnessed neural networks to reveal that drugs and their doses (inputs) can be correlated to quantifiable measures of treatment efficacy and safety (outputs) through a parabolic response surface (PRS), which is governed by a second order polynomial . Importantly, unlike most conventional approaches which often involve fixed or static magnitudes of treatment, AI‐PRS is a disease indication‐agnostic approach that has been subsequently used to optimize both drug and dose selection as well as dynamic dosing from the preclinical through clinical stages of development without the need for complex disease mechanism data . More specifically, the PRS approach can be described using the following equation

V false(C false) = x_{0} + \sum x_{i} C_{i} + \sum y_{i} C_{i}^{2} + \sum z_{i j} C_{i} C_{j}

V ( C ) is the efficacy, or viral load in this study, C i is the dose of drug i , and x 0 , x i , y i , and z ij are patient‐specific constants that are determined during the clinical calibration process .…”

Section: Introductionmentioning

confidence: 99%

Harnessing Artificial Intelligence to Optimize Long‐Term Maintenance Dosing for Antiretroviral‐Naive Adults with HIV‐1 Infection

Shen

Liu

et al. 2019

Advanced Therapeutics

View full text Add to dashboard Cite

Antiretroviral therapy (ART) serves as a mainstay in treating human immunodeficiency virus (HIV)infection. An HIV patient is traditionally administered the same ART regimen for life, even if his/her viral load has been reduced by several orders of magnitude from the initial viral load. Dose reduction in ART has been clinically explored in a trial-and-error manner to reduce side effects and improve ART sustainability. Using artificial intelligence (AI), we have discovered that drugs and doses inputs can be related to viral load reduction through a Parabolic Response Surface (PRS). The AI-PRS platform can rationally guide a clinically-actionable approach to identify optimized population-wide and personalized dosing. In this prospective pilot clinical trial, a combination regimen of tenofovir (TDF), efavirenz (EFV) and lamivudine (3TC) is administered to ten patients. Using AI-PRS, a 33% reduction in the long-term TDF maintenance dose (200 mg) is identified compared to standard regimens (300 mg). This regimen keeps the HIV viral load below 40 copies/mL with no relapse during a 144-week observation period. This study demonstrates that AI-PRS can potentially serve as a scalable approach to optimize and sustain the long-term management of HIV as well as a broad spectrum of other indications.

show abstract

“…This is due to the fact that drug synergy is dose dependent and as such, delivering what are perceived to be suitable drugs at the wrong dose ratios can lead to suboptimal responses . Promising strategies to design novel drug combinations have included pairwise drug predictions, systems biology‐guided drug combination design, as well as ex vivo and disease modeling approaches, among others, in an effort to estimate how diseased systems will respond to multi‐drug inputs . Furthermore, delivering drugs that may have never been considered at the right dose ratios can lead to treatment responses that markedly exceed conventional regimens .…”

Section: Introductionmentioning

confidence: 99%

Harnessing an Artificial Intelligence Platform to Dynamically Individualize Combination Therapy for Treating Colorectal Carcinoma in a Rat Model

Chang

et al. 2019

Advanced Therapeutics

View full text Add to dashboard Cite

Designing multi-drug regimens often involves target-and synergy prediction-based drug selection, and subsequent dose escalation to achieve the maximum tolerated dose (MTD) of each drug. This approach may improve efficacy, but not the optimal efficacy and often substantially increases toxicity. Drug interactions depend on many pathways in the omics networks and further complicate the design process. The virtually infinite drug-dose parameter space cannot be reconciled using conventional approaches, which are largely based on prediction. This barrier at least partially accounts for the low response rates that are observed with conventional mono-and combinatorial chemotherapy. A combination of nonstandard therapies for colorectal cancer (AGCH: adriamycin, gemcitabine, cisplatin, and herceptin) at ¼ MTD is used to treat the rats, the tumor response rates varied in a wide range. Some of the tumor response rates are close to that of control group. This work harnesses an artificial intelligence (AI) platform that is mechanism free and can dynamically optimize combinatorial therapy in rats. The individually optimized AGCH regimen reveal starkly different drug-dose parameters, which can converge each rat toward the same and low tumor response rate. Importantly, this AI-based drug-dose optimization technology is an actionable platform, which can achieve N-of-1 therapy.

show abstract

Prediction of ultra-high-order antibiotic combinations based on pairwise interactions

Cited by 59 publications

References 34 publications

Weakest-Link Dynamics Predict Apparent Antibiotic Interactions in a Model Cross-Feeding Community

Weakest-Link Dynamics Predict Apparent Antibiotic Interactions in a Model Cross-Feeding Community

Harnessing Artificial Intelligence to Optimize Long‐Term Maintenance Dosing for Antiretroviral‐Naive Adults with HIV‐1 Infection

Harnessing an Artificial Intelligence Platform to Dynamically Individualize Combination Therapy for Treating Colorectal Carcinoma in a Rat Model

Contact Info

Product

Resources

About