As an approach to personalized medicine, the study proposes that "mutational fingerprints" of DNA repair are a promising predictive genetic marker to predict which tumors will respond to certain therapies.
The results have been published in Nature Communications.

Cancer treatment is increasingly moving towards a personalized approach, in which genetic changes in a tumor can be used to determine the best therapeutic strategy to treat it. Until now, in many cases, the genetic changes included what is known as a cancer “driver” mutation, which predicted response to a drug. For example, mutations of the BRAF gene in melanoma predict response to inhibitor drugs. BRAF, and ERBB2 gene amplifications in breast cancer predict response to ERBB-inhibiting drugs.
However, these examples of successful pharmacological markers are still rare. For many mutated “driver” genes, specific drugs targeting them are not available. In addition, tumors from different patients show great variability in response to drugs, and this variability is not usually linked to mutations in these driver genes.
Scientists at IRB Barcelona, led by Dr. Fran Supek, ICREA researcher and head of the Genome Data Science laboratory, have discovered that so-called "mutational signatures" can accurately predict the response to various drugs of cancer cells from many types of tumors. These "mutational signatures" do not focus on driver genes, but rather represent a set of mutations found throughout the tumor genome. Mutational signatures may reflect, for example, that the tumor has difficulty copying or repairing DNA, which may make it more amenable to treatment.
"We have performed a statistical analysis with machine learning methods, jointly considering the genomes of cancer cells, their response to various drugs, and their response to gene editing experiments. Surprisingly, our analysis revealed that the ‘classical’ genetic markers, such as driver gene mutations or copy number changes, often have less potential than the mutational signature for predicting drug response," explains Dr. Supek.
Deficiencies in DNA repair make cancer cells easier to target by many drugs
This study found many statistical predictions linking a particular mutational signature to tumor response to a drug. It was already known that a certain type of deficiency in the BRCA genes – which can cause cancers of the breast, ovary and prostate – predicts the response to drugs that target BRCA deficiency. This deficiency, in turn, leaves a mutational signature in the genome in the form of the deletion (or removal) of some DNA fragments, which may signal that the tumor can be treated with drugs targeting BRCA deficiency.
In this work, led by postdoctoral researcher Dr. Jurica Levatić, now at the Jozef Stefan Institute in Slovenia, it has been shown that this is just one example of many. Several types of DNA repair deficiency, such as defects in "spell check," which detects errors in the DNA copying process, can make cancer cells more vulnerable to certain drugs. Since tumors have impaired DNA repair mechanisms, these planned therapies would have a greater ability to eliminate cancer cells and preserve healthy ones.
Previous exposure to mutagenic chemicals, including drugs, can confer resistance to future therapies in cancer cells
The statistical and machine learning analyzes in this work, jointly implemented by Marina Salvadores, a Ph.D. student in the Genome Data Science lab, can connect databases from previous experiments in which many drugs had been tested on cancer cells growing in vitro (in the laboratory). In addition, this study also integrated experimental "gene editing" data, in which the CRISPR technique was used to turn off multiple drug target genes in the same types of cancer cells. This approach enabled the researchers to link drug target genes to response to drug treatments, thus reinforcing the finding that mutational signatures are a good predictor of drug response in cancer.
Interestingly, cancer cells that carried genomic "scars" (mutational signatures) from previous exposure to mutagenic chemicals tended to be resistant to various drugs. A possible explanation for this is based on the known mechanism by which, for example, brain cancer cells can deactivate their DNA repair systems during treatment with the mutagenic drug TMZ, which could permanently convert them into cancer cells. resistant and hypermutant to a series of future treatments.
The study is suggests that this type of adaptation may be common in cancer. This has potential implications, as tumors caused by exposure to mutagens, for example, lung exposure to tobacco or skin exposure to ultraviolet light, may be more difficult to treat, since the cells may harbor a " long-term memory" to cope with DNA damage.
The algorithms used to identify mutational signatures and link them to drug vulnerabilities are open access. Future work from the lab will focus on testing these prediction algorithms on patient data, thus overcoming the challenge of the paucity of public patient genomic data that correlates with randomized clinical trials.
This work has been funded by the EU through the ERC Starting Grant "HYPER-INSIGHT", the "DECIDER" project of RIA Horizon 2020, and the Spanish Ministry of Science and Innovation.
Cancer treatment is increasingly moving towards a personalized approach, in which genetic changes in a tumor can be used to determine the best therapeutic strategy to treat it. Until now, in many cases, the genetic changes included what is known as a cancer “driver” mutation, which predicted response to a drug. For example, mutations of the BRAF gene in melanoma predict response to inhibitor drugs. BRAF, and ERBB2 gene amplifications in breast cancer predict response to ERBB-inhibiting drugs.
However, these examples of successful pharmacological markers are still rare. For many mutated “driver” genes, specific drugs targeting them are not available. In addition, tumors from different patients show great variability in response to drugs, and this variability is not usually linked to mutations in these driver genes.
Scientists at IRB Barcelona, led by Dr. Fran Supek, ICREA researcher and head of the Genome Data Science laboratory, have discovered that so-called "mutational signatures" can accurately predict the response to various drugs of cancer cells from many types of tumors. These "mutational signatures" do not focus on driver genes, but rather represent a set of mutations found throughout the tumor genome. Mutational signatures may reflect, for example, that the tumor has difficulty copying or repairing DNA, which may make it more amenable to treatment.
"We have performed a statistical analysis with machine learning methods, jointly considering the genomes of cancer cells, their response to various drugs, and their response to gene editing experiments. Surprisingly, our analysis revealed that the ‘classical’ genetic markers, such as driver gene mutations or copy number changes, often have less potential than the mutational signature for predicting drug response," explains Dr. Supek.
Deficiencies in DNA repair make cancer cells easier to target by many drugs
This study found many statistical predictions linking a particular mutational signature to tumor response to a drug. It was already known that a certain type of deficiency in the BRCA genes – which can cause cancers of the breast, ovary and prostate – predicts the response to drugs that target BRCA deficiency. This deficiency, in turn, leaves a mutational signature in the genome in the form of the deletion (or removal) of some DNA fragments, which may signal that the tumor can be treated with drugs targeting BRCA deficiency.
In this work, led by postdoctoral researcher Dr. Jurica Levatić, now at the Jozef Stefan Institute in Slovenia, it has been shown that this is just one example of many. Several types of DNA repair deficiency, such as defects in "spell check," which detects errors in the DNA copying process, can make cancer cells more vulnerable to certain drugs. Since tumors have impaired DNA repair mechanisms, these planned therapies would have a greater ability to eliminate cancer cells and preserve healthy ones.
Previous exposure to mutagenic chemicals, including drugs, can confer resistance to future therapies in cancer cells
The statistical and machine learning analyzes in this work, jointly implemented by Marina Salvadores, a Ph.D. student in the Genome Data Science lab, can connect databases from previous experiments in which many drugs had been tested on cancer cells growing in vitro (in the laboratory). In addition, this study also integrated experimental "gene editing" data, in which the CRISPR technique was used to turn off multiple drug target genes in the same types of cancer cells. This approach enabled the researchers to link drug target genes to response to drug treatments, thus reinforcing the finding that mutational signatures are a good predictor of drug response in cancer.
Interestingly, cancer cells that carried genomic "scars" (mutational signatures) from previous exposure to mutagenic chemicals tended to be resistant to various drugs. A possible explanation for this is based on the known mechanism by which, for example, brain cancer cells can deactivate their DNA repair systems during treatment with the mutagenic drug TMZ, which could permanently convert them into cancer cells. resistant and hypermutant to a series of future treatments.
The study is suggests that this type of adaptation may be common in cancer. This has potential implications, as tumors caused by exposure to mutagens, for example, lung exposure to tobacco or skin exposure to ultraviolet light, may be more difficult to treat, since the cells may harbor a " long-term memory" to cope with DNA damage.
The algorithms used to identify mutational signatures and link them to drug vulnerabilities are open access. Future work from the lab will focus on testing these prediction algorithms on patient data, thus overcoming the challenge of the paucity of public patient genomic data that correlates with randomized clinical trials.
This work has been funded by the EU through the ERC Starting Grant "HYPER-INSIGHT", the "DECIDER" project of RIA Horizon 2020, and the Spanish Ministry of Science and Innovation.
Responsible

Juan Bautista Barroso Albarracín
Catedrático UJA

Juan Bautista Barroso Albarracín
Catedrático UJA
- Email:info@example.com
Participants

Roberto García Ruiz
Catedrático UJA

Roberto García Ruiz
Catedrático UJA
- Email:info@example.com

Miguel Delgado Rodríguez
Catedrático UJA

Miguel Delgado Rodríguez
Catedrático UJA
- Email:info@example.com

Francisco Luque Vázquez
Catedrático UJA

Francisco Luque Vázquez
Catedrático UJA
- Email:info@example.com

Sebastián Sánchez Villasclaras
Catedrático UJA

Sebastián Sánchez Villasclaras
Catedrático UJA
- Email:info@example.com

Francisco José Torres Ruiz
Catedrático UJA

Francisco José Torres Ruiz
Catedrático UJA
- Email:info@example.com

Antonio Molina Díaz
Catedrático UJA

Antonio Molina Díaz
Catedrático UJA
- Email:info@example.com

Juan Gómez Ortega
Catedrático UJA

Juan Gómez Ortega
Catedrático UJA
- Email:info@example.com

José Juan Gaforio Martínez
Catedrático UJA

José Juan Gaforio Martínez
Catedrático UJA
- Email:info@example.com