User login
Researchers say they have found ways to identify patients with a high risk of developing acute myeloid leukemia (AML).
The researchers used basic clinical and laboratory data to identify patients 6 to 12 months before AML presentation.
Using genetic information, the researchers were able to identify high-risk patients several years before AML presentation.
“This long time window gives us the first opportunity to think about how to prevent AML,” said John Dick, PhD, of Princess Margaret Cancer Centre in Toronto, Ontario, Canada.
However, Dr Dick and his colleagues noted that neither the genetic method nor the clinical data method were entirely accurate in identifying patients who would develop AML.
The researchers described both methods in Nature.
The team’s work began with the goal of distinguishing patients who will develop AML from patients who simply develop age-related clonal hematopoiesis.
“We wanted to know if there was any difference between these 2 groups in the genetics of their ‘normal’ blood samples taken at enrollment,” Dr Dick said. “To find out, we developed a gene-sequencing tool that captured the most common genes that get altered in AML and sequenced all the 500 blood samples.”
The researchers analyzed samples from 95 patients who ultimately developed AML and 414 age- and gender-matched controls. The samples from AML patients were obtained an average of 6.3 years before AML diagnosis.
The researchers found that pre-AML cases had more mutations per sample, higher variant allele frequencies, and a higher frequency of certain mutations than controls. Specifically, pre-AML cases had a significantly greater frequency of DNMT3A, TET2, SRSF2, ASXL1, TP53, U2AF1, JAK2, RUNX1, and IDH2 mutations.
The researchers used these findings to develop a model for predicting AML. They tested the model in a validation cohort of 29 pre-AML cases and 262 controls, as well as in a cohort combining the validation and discovery cohorts. The model predicted AML development with 41.9% sensitivity and 95.7% specificity.
However, the researchers said widespread use of this model would not be practical because AML is so rare. The team said millions of people would need to undergo screening to identify a few pre-AML cases, and there would be “many” false-positives.
Therefore, the researchers developed a model using clinical information from electronic health records—particularly blood counts. This model was able to predict AML development 6 to 12 months before diagnosis with a sensitivity of 25.7% and specificity of 98.2%.
The researchers said these results suggest clinical data can be used to identify patients with a high risk of AML who may benefit from targeted genetic screening. And combining clinical and genetic information in a single model could improve predictive accuracy.
“Our study provides, for the first time, evidence that we can identify people at risk of developing AML many years before they actually develop this life-threatening disease,” said study author George Vassiliou, PhD, of the Wellcome Trust Sanger Institute in Hinxton, UK.
“We hope to build on these findings to develop robust screening tests for identifying those at risk and drive research into how to prevent or stall progression towards AML. Our aspiration is that, one day, AML prevention would provide a compelling alternative to treatment.”
Researchers say they have found ways to identify patients with a high risk of developing acute myeloid leukemia (AML).
The researchers used basic clinical and laboratory data to identify patients 6 to 12 months before AML presentation.
Using genetic information, the researchers were able to identify high-risk patients several years before AML presentation.
“This long time window gives us the first opportunity to think about how to prevent AML,” said John Dick, PhD, of Princess Margaret Cancer Centre in Toronto, Ontario, Canada.
However, Dr Dick and his colleagues noted that neither the genetic method nor the clinical data method were entirely accurate in identifying patients who would develop AML.
The researchers described both methods in Nature.
The team’s work began with the goal of distinguishing patients who will develop AML from patients who simply develop age-related clonal hematopoiesis.
“We wanted to know if there was any difference between these 2 groups in the genetics of their ‘normal’ blood samples taken at enrollment,” Dr Dick said. “To find out, we developed a gene-sequencing tool that captured the most common genes that get altered in AML and sequenced all the 500 blood samples.”
The researchers analyzed samples from 95 patients who ultimately developed AML and 414 age- and gender-matched controls. The samples from AML patients were obtained an average of 6.3 years before AML diagnosis.
The researchers found that pre-AML cases had more mutations per sample, higher variant allele frequencies, and a higher frequency of certain mutations than controls. Specifically, pre-AML cases had a significantly greater frequency of DNMT3A, TET2, SRSF2, ASXL1, TP53, U2AF1, JAK2, RUNX1, and IDH2 mutations.
The researchers used these findings to develop a model for predicting AML. They tested the model in a validation cohort of 29 pre-AML cases and 262 controls, as well as in a cohort combining the validation and discovery cohorts. The model predicted AML development with 41.9% sensitivity and 95.7% specificity.
However, the researchers said widespread use of this model would not be practical because AML is so rare. The team said millions of people would need to undergo screening to identify a few pre-AML cases, and there would be “many” false-positives.
Therefore, the researchers developed a model using clinical information from electronic health records—particularly blood counts. This model was able to predict AML development 6 to 12 months before diagnosis with a sensitivity of 25.7% and specificity of 98.2%.
The researchers said these results suggest clinical data can be used to identify patients with a high risk of AML who may benefit from targeted genetic screening. And combining clinical and genetic information in a single model could improve predictive accuracy.
“Our study provides, for the first time, evidence that we can identify people at risk of developing AML many years before they actually develop this life-threatening disease,” said study author George Vassiliou, PhD, of the Wellcome Trust Sanger Institute in Hinxton, UK.
“We hope to build on these findings to develop robust screening tests for identifying those at risk and drive research into how to prevent or stall progression towards AML. Our aspiration is that, one day, AML prevention would provide a compelling alternative to treatment.”
Researchers say they have found ways to identify patients with a high risk of developing acute myeloid leukemia (AML).
The researchers used basic clinical and laboratory data to identify patients 6 to 12 months before AML presentation.
Using genetic information, the researchers were able to identify high-risk patients several years before AML presentation.
“This long time window gives us the first opportunity to think about how to prevent AML,” said John Dick, PhD, of Princess Margaret Cancer Centre in Toronto, Ontario, Canada.
However, Dr Dick and his colleagues noted that neither the genetic method nor the clinical data method were entirely accurate in identifying patients who would develop AML.
The researchers described both methods in Nature.
The team’s work began with the goal of distinguishing patients who will develop AML from patients who simply develop age-related clonal hematopoiesis.
“We wanted to know if there was any difference between these 2 groups in the genetics of their ‘normal’ blood samples taken at enrollment,” Dr Dick said. “To find out, we developed a gene-sequencing tool that captured the most common genes that get altered in AML and sequenced all the 500 blood samples.”
The researchers analyzed samples from 95 patients who ultimately developed AML and 414 age- and gender-matched controls. The samples from AML patients were obtained an average of 6.3 years before AML diagnosis.
The researchers found that pre-AML cases had more mutations per sample, higher variant allele frequencies, and a higher frequency of certain mutations than controls. Specifically, pre-AML cases had a significantly greater frequency of DNMT3A, TET2, SRSF2, ASXL1, TP53, U2AF1, JAK2, RUNX1, and IDH2 mutations.
The researchers used these findings to develop a model for predicting AML. They tested the model in a validation cohort of 29 pre-AML cases and 262 controls, as well as in a cohort combining the validation and discovery cohorts. The model predicted AML development with 41.9% sensitivity and 95.7% specificity.
However, the researchers said widespread use of this model would not be practical because AML is so rare. The team said millions of people would need to undergo screening to identify a few pre-AML cases, and there would be “many” false-positives.
Therefore, the researchers developed a model using clinical information from electronic health records—particularly blood counts. This model was able to predict AML development 6 to 12 months before diagnosis with a sensitivity of 25.7% and specificity of 98.2%.
The researchers said these results suggest clinical data can be used to identify patients with a high risk of AML who may benefit from targeted genetic screening. And combining clinical and genetic information in a single model could improve predictive accuracy.
“Our study provides, for the first time, evidence that we can identify people at risk of developing AML many years before they actually develop this life-threatening disease,” said study author George Vassiliou, PhD, of the Wellcome Trust Sanger Institute in Hinxton, UK.
“We hope to build on these findings to develop robust screening tests for identifying those at risk and drive research into how to prevent or stall progression towards AML. Our aspiration is that, one day, AML prevention would provide a compelling alternative to treatment.”