- Recurrent protein expression patterns in pediatric AML are independently prognostic of patient outcomes.
- Certain protein expression patterns are associated with better responses to given therapies.
- Combining proteomic data with genetic risk-stratification data may help tailor and individualize therapy for patients with pediatric AML.
Protein expression patterns in patients with pediatric acute myeloid leukemia (AML) are independently prognostic of survival outcomes and responses to certain treatments, according to a recent study.
Fieke W. Hoff, MD, PhD, of the University Medical Center Groningen at the University of Groningen in the Netherlands and the University of Texas Southwestern Medical Center, and colleagues conducted the study and published its findings in Haematologica.
Dr. Hoff and colleagues studied the “proteomic landscape” of pediatric AML to identify potential therapeutic targets because “with the exception of acute promyelocytic leukemia, leukemia with FLT3-internal tandem duplication mutations, and mixed phenotype acute leukemia, pediatric AML has been treated as a homogeneous disease, as therapy does not differ based on the underlying mutations.”
The researchers previously conducted a pilot study that assessed 194 proteins in 95 patients with de novo pediatric AML, identifying eight prognostic protein expression signatures.
Dr. Hoff and colleagues used the same approach in the current study to prospectively assess samples from patients treated in a Children’s Oncology Group randomized, phase III trial that evaluated the impact of adding the proteasome inhibitor bortezomib to cytarabine, daunorubicin, and etoposide.
The current study aimed to “validate the ability to classify pediatric AML patients based on proteomics in a larger cohort, with significantly more protein targets,” according to the investigators. They also wanted to determine if protein classification can enhance risk stratification; identify patients who could benefit from bortezomib, cytarabine, daunorubicin, and etoposide; and identify additional potential therapeutic targets.
The researchers collected peripheral blood samples from 500 patients with de novo pediatric AML, as well as 30 control CD34-positive bone marrow samples from 20 children and 10 adults between July 2011 and February 2017. They collected samples from all patients prior to chemotherapy.
The researchers were able to obtain samples from 92.6% of patients 10 hours after the start of induction chemotherapy and from 93.2% of patients 24 hours after the start of chemotherapy. Patients received one dose of each chemotherapeutic agent by the time samples were collected 10 hours and 24 hours after therapy.
The researchers used 296 “strictly validated” antibodies on reverse phase protein arrays to measure protein expression and activation levels in samples from patients with pediatric AML and control samples. They used the multistep MetaGalaxy analysis methodology to identify protein expression signatures “based on strong recurrent protein expression patterns,” according to Dr. Hoff and colleagues.
The study’s investigators analyzed 296 proteins and allocated them into 31 protein functional groups, with three to five protein clusters identified in each group for a total of 116 protein clusters. The researchers applied principal component analysis to graphically compare patient protein cluster expression patterns with patterns in non-malignant CD34-positive cells.
The overall proteomic profiles of patients with pediatric AML were “distinct from those of normal CD34-positive cells,” but “overlapping ‘normal-like’ expression patterns” occurred in 27% of the protein clusters, the study’s authors wrote. Protein clusters that did not have dominant co-localization on CD34-positive samples were defined as leukemia-specific clusters.
They determined seven (23%) of the protein functional groups were significantly associated with patient outcomes. One example is the heat shock protein functional group, which was divided into four protein clusters: protein cluster one, protein cluster two, protein cluster three, and protein cluster four. The heat shock protein clusters were significantly prognostic in all patients for overall survival (OS; P=.004), as well as for event-free survival (EFS; P=.0009) and relapse risk (P=.0016).
The heat shock protein expression cluster patterns were also prognostic for OS, EFS, and relapse risk in patients receiving bortezomib plus cytarabine, daunorubicin, and etoposide, or cytarabine, daunorubicin, and etoposide.
Patients with protein cluster two who received the quadruplet had a significantly higher five-year OS rate (81%) than those who received the triplet (54%; P=.00087). However, patients with protein cluster four who received the triplet had a five-year OS rate of 100% (see TABLE 1), while those who received the quadruplet had a five-year OS rate of 67% (P=.019). Bortezomib did not affect patients with protein cluster one, “which was an unfavorable prognostic indicator” after treatment with the triplet or quadruplet, the authors wrote.
The researchers also used the 116 protein clusters identified to derive correlated protein clusters, defined as protein constellations. They then defined nine protein expression signatures “as clusters of patients who expressed similar combinations of [protein constellations],” according to Dr. Hoff and colleagues. They found 24 proteins were universally altered across all protein expression signatures, with seven proteins having universally higher expression and 17 having universally lower expression.
The protein expression signatures were associated with cytogenetics, mutation status, and patient prognosis. See TABLE 2 for a summary of the data on protein expression signatures and prognosis. In a multivariate Cox regression analysis, unfavorable protein expression signatures remained an independent prognostic factor for OS, EFS, and relapse risk.
Dr. Hoff and colleagues also identified protein expression signatures that were associated with patient responses to therapy. In patients receiving the triplet, protein expression signatures five through eight were associated with the poorest prognosis. Adding bortezomib to the triplet treatment improved five-year OS rates in patients with protein expression signatures six, seven, and eight.
Protein expression signature risk groups did not correlate with the risk group stratification used in the clinical trial or conventional risk group stratification. However, protein expression signatures stratified EFS and relapse risk in patients classified in the trial as low risk, defined by inv(16)/t(16;16), t(8;21), NPM1, or CEBPA mutations. Protein expression signatures one, three, six, and nine were associated with favorable prognosis, while signatures five and seven were associated with unfavorable prognosis in the low-risk patients.
Protein expression signature risk groups were also prognostic for patients who were classified as high risk in the trial, and signatures one, three, and six were associated with a favorable prognosis. However, protein expression signature nine, which was associated with a favorable prognosis in low-risk patients, was associated with a “highly unfavorable” prognosis in high-risk patients, the researchers wrote.
Proteomics ‘Can Direct Therapy’ in Pediatric AML
The research, which was the “largest proteomic study in pediatric AML” to the authors’ knowledge, indicated the “genetic heterogeneity of pediatric AML coalesces into a finite number of recurrent protein expression patterns,” Dr. Hoff and colleagues wrote.
It showed protein expression signatures can be significantly prognostic, especially combined with genetic data, “demonstrating that adding proteomics to genetic risk-stratification can direct therapy leading to improved outcome,” according to the researchers.
The prognostic proteomic data may help clinicians and researchers tailor the selection and development of therapies for pediatric AML.
“In summary, we confirmed the existence of recurrent protein patterns in pediatric AML, which enabled separation of AML patients into recurrent [protein expression signatures] that were prognostic, particularly when combined with known pediatric AML risk factors,” Dr. Hoff and colleagues concluded. “We identified [protein expression signatures] that benefited from [bortezomib, cytarabine, daunorubicin, and etoposide] and postulate that recognition of abnormal proteins can aid in risk stratification and therapy selection in pediatric, and perhaps adult, AML.”
The study was funded in part by Takeda Pharmaceuticals.
Hoff FW, Van Dijk AD, Qiu Y, et al. Clinical relevance of proteomic profiling in de novo pediatric acute myeloid leukemia: a Children’s Oncology Group study. Haematologica. 2022;107(10):2329-2343. doi:10.3324/haematol.2021.279672