Limits...
Cell signaling-based classifier predicts response to induction therapy in elderly patients with acute myeloid leukemia.

Cesano A, Willman CL, Kopecky KJ, Gayko U, Putta S, Louie B, Westfall M, Purvis N, Spellmeyer DC, Marimpietri C, Cohen AC, Hackett J, Shi J, Walker MG, Sun Z, Paietta E, Tallman MS, Cripe LD, Atwater S, Appelbaum FR, Radich JP - PLoS ONE (2015)

Bottom Line: Importantly, a classifier developed using only clinical and molecular inputs from the same sample set (DXCLINICAL2) lacked prediction accuracy: AUROC = 0.61 (p = 0.18) in the BM Verification Set and 0.53 (p = 0.38) in the BM Validation Set.Notably, the DXSCNP classifier was still significant in predicting response in the BM Validation Analysis Set after controlling for DXCLINICAL2 (p = 0.03), showing that DXSCNP provides information that is independent from that provided by currently used prognostic markers.Taken together, these data show that the proteomic classifier may provide prognostic information relevant to treatment planning beyond genetic mutations and traditional prognostic factors in elderly AML.

View Article: PubMed Central - PubMed

Affiliation: Nodality, Inc., South San Francisco, California, United States of America.

ABSTRACT
Single-cell network profiling (SCNP) data generated from multi-parametric flow cytometry analysis of bone marrow (BM) and peripheral blood (PB) samples collected from patients >55 years old with non-M3 AML were used to train and validate a diagnostic classifier (DXSCNP) for predicting response to standard induction chemotherapy (complete response [CR] or CR with incomplete hematologic recovery [CRi] versus resistant disease [RD]). SCNP-evaluable patients from four SWOG AML trials were randomized between Training (N = 74 patients with CR, CRi or RD; BM set = 43; PB set = 57) and Validation Analysis Sets (N = 71; BM set = 42, PB set = 53). Cell survival, differentiation, and apoptosis pathway signaling were used as potential inputs for DXSCNP. Five DXSCNP classifiers were developed on the SWOG Training set and tested for prediction accuracy in an independent BM verification sample set (N = 24) from ECOG AML trials to select the final classifier, which was a significant predictor of CR/CRi (area under the receiver operating characteristic curve AUROC = 0.76, p = 0.01). The selected classifier was then validated in the SWOG BM Validation Set (AUROC = 0.72, p = 0.02). Importantly, a classifier developed using only clinical and molecular inputs from the same sample set (DXCLINICAL2) lacked prediction accuracy: AUROC = 0.61 (p = 0.18) in the BM Verification Set and 0.53 (p = 0.38) in the BM Validation Set. Notably, the DXSCNP classifier was still significant in predicting response in the BM Validation Analysis Set after controlling for DXCLINICAL2 (p = 0.03), showing that DXSCNP provides information that is independent from that provided by currently used prognostic markers. Taken together, these data show that the proteomic classifier may provide prognostic information relevant to treatment planning beyond genetic mutations and traditional prognostic factors in elderly AML.

Show MeSH
Study Design Diagram.Flowchart of the study design with descriptive schematics of the patient sets in the Training, Verification and Validation analysis sets. SWOG samples were randomized into a Training and a Validation Analysis set and were sorted by tissue type (PB or BM). An initial subset of classifiers was trained separately in PB and BM samples in the Training Analysis sets and then PB classifiers were applied to BM and BM classifiers were applied to PB. From this training process 5 candidate classifiers were selected and applied to the ECOG Verification Analysis set. The final SCNP classifier was further refined and applied to 1) ECOG Verification Analysis set, 2) SWOG BM Validation Analysis set and 3) SWOG PB Validation Analysis set.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4401549&req=5

pone.0118485.g004: Study Design Diagram.Flowchart of the study design with descriptive schematics of the patient sets in the Training, Verification and Validation analysis sets. SWOG samples were randomized into a Training and a Validation Analysis set and were sorted by tissue type (PB or BM). An initial subset of classifiers was trained separately in PB and BM samples in the Training Analysis sets and then PB classifiers were applied to BM and BM classifiers were applied to PB. From this training process 5 candidate classifiers were selected and applied to the ECOG Verification Analysis set. The final SCNP classifier was further refined and applied to 1) ECOG Verification Analysis set, 2) SWOG BM Validation Analysis set and 3) SWOG PB Validation Analysis set.

Mentions: SCNP assays for all SWOG patient samples were performed blindly to all clinical data and in a random order as part of a single experiment. Patients with assessable BM and/or PB SCNP results (n = 213) were then randomized approximately 1:1 to Training and Validation Sets (see Figs 2 and 4). The minimization approach of Pocock and Simon [17] was used to balance disease characteristics and other relevant variables between the Training and Validation Sets. These included: response to induction therapy, sample type(s), cytogenetic risk group assigned per SWOG protocol, SWOG parent trial treatment arm, and FLT3-ITD mutation in BM and/or PB samples, and extent of proteomic readout availability in BM and/or PB samples (see S1 Methods). Within the Training and Validation Sets, only patients having an induction outcome of CR, CRi, or RD were assigned to the Training and Validation Analysis Sets; patients with TRM were excluded from the Analysis Sets since the assay was specifically designed to measure blast chemosensitivity and not comorbidities [9], [18] (Fig 2). Clinical and molecular variables from 74 patients randomized to the Training Set were used to develop DXCLINICAL1 and DXCLINICAL2. SCNP data for the BM (n = 43) and PB (n = 57) Training Analysis Sets were used to develop the SCNP-based predictive models. Since some patients had two SCNP-assessable samples (BM and PB), a partial overlap existed between the patients in the BM and PB Analysis Sets. However, within each Analysis Set, each patient contributed only one sample (i.e., only one tissue type) (Fig 4). Values of inputs for DXCLINICAL1 and DXCLINICAL2 which were missing (≤6% for any input except for cytogenetics which was missing in approximately 20%), were imputed as described in S3 Methods.


Cell signaling-based classifier predicts response to induction therapy in elderly patients with acute myeloid leukemia.

Cesano A, Willman CL, Kopecky KJ, Gayko U, Putta S, Louie B, Westfall M, Purvis N, Spellmeyer DC, Marimpietri C, Cohen AC, Hackett J, Shi J, Walker MG, Sun Z, Paietta E, Tallman MS, Cripe LD, Atwater S, Appelbaum FR, Radich JP - PLoS ONE (2015)

Study Design Diagram.Flowchart of the study design with descriptive schematics of the patient sets in the Training, Verification and Validation analysis sets. SWOG samples were randomized into a Training and a Validation Analysis set and were sorted by tissue type (PB or BM). An initial subset of classifiers was trained separately in PB and BM samples in the Training Analysis sets and then PB classifiers were applied to BM and BM classifiers were applied to PB. From this training process 5 candidate classifiers were selected and applied to the ECOG Verification Analysis set. The final SCNP classifier was further refined and applied to 1) ECOG Verification Analysis set, 2) SWOG BM Validation Analysis set and 3) SWOG PB Validation Analysis set.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4401549&req=5

pone.0118485.g004: Study Design Diagram.Flowchart of the study design with descriptive schematics of the patient sets in the Training, Verification and Validation analysis sets. SWOG samples were randomized into a Training and a Validation Analysis set and were sorted by tissue type (PB or BM). An initial subset of classifiers was trained separately in PB and BM samples in the Training Analysis sets and then PB classifiers were applied to BM and BM classifiers were applied to PB. From this training process 5 candidate classifiers were selected and applied to the ECOG Verification Analysis set. The final SCNP classifier was further refined and applied to 1) ECOG Verification Analysis set, 2) SWOG BM Validation Analysis set and 3) SWOG PB Validation Analysis set.
Mentions: SCNP assays for all SWOG patient samples were performed blindly to all clinical data and in a random order as part of a single experiment. Patients with assessable BM and/or PB SCNP results (n = 213) were then randomized approximately 1:1 to Training and Validation Sets (see Figs 2 and 4). The minimization approach of Pocock and Simon [17] was used to balance disease characteristics and other relevant variables between the Training and Validation Sets. These included: response to induction therapy, sample type(s), cytogenetic risk group assigned per SWOG protocol, SWOG parent trial treatment arm, and FLT3-ITD mutation in BM and/or PB samples, and extent of proteomic readout availability in BM and/or PB samples (see S1 Methods). Within the Training and Validation Sets, only patients having an induction outcome of CR, CRi, or RD were assigned to the Training and Validation Analysis Sets; patients with TRM were excluded from the Analysis Sets since the assay was specifically designed to measure blast chemosensitivity and not comorbidities [9], [18] (Fig 2). Clinical and molecular variables from 74 patients randomized to the Training Set were used to develop DXCLINICAL1 and DXCLINICAL2. SCNP data for the BM (n = 43) and PB (n = 57) Training Analysis Sets were used to develop the SCNP-based predictive models. Since some patients had two SCNP-assessable samples (BM and PB), a partial overlap existed between the patients in the BM and PB Analysis Sets. However, within each Analysis Set, each patient contributed only one sample (i.e., only one tissue type) (Fig 4). Values of inputs for DXCLINICAL1 and DXCLINICAL2 which were missing (≤6% for any input except for cytogenetics which was missing in approximately 20%), were imputed as described in S3 Methods.

Bottom Line: Importantly, a classifier developed using only clinical and molecular inputs from the same sample set (DXCLINICAL2) lacked prediction accuracy: AUROC = 0.61 (p = 0.18) in the BM Verification Set and 0.53 (p = 0.38) in the BM Validation Set.Notably, the DXSCNP classifier was still significant in predicting response in the BM Validation Analysis Set after controlling for DXCLINICAL2 (p = 0.03), showing that DXSCNP provides information that is independent from that provided by currently used prognostic markers.Taken together, these data show that the proteomic classifier may provide prognostic information relevant to treatment planning beyond genetic mutations and traditional prognostic factors in elderly AML.

View Article: PubMed Central - PubMed

Affiliation: Nodality, Inc., South San Francisco, California, United States of America.

ABSTRACT
Single-cell network profiling (SCNP) data generated from multi-parametric flow cytometry analysis of bone marrow (BM) and peripheral blood (PB) samples collected from patients >55 years old with non-M3 AML were used to train and validate a diagnostic classifier (DXSCNP) for predicting response to standard induction chemotherapy (complete response [CR] or CR with incomplete hematologic recovery [CRi] versus resistant disease [RD]). SCNP-evaluable patients from four SWOG AML trials were randomized between Training (N = 74 patients with CR, CRi or RD; BM set = 43; PB set = 57) and Validation Analysis Sets (N = 71; BM set = 42, PB set = 53). Cell survival, differentiation, and apoptosis pathway signaling were used as potential inputs for DXSCNP. Five DXSCNP classifiers were developed on the SWOG Training set and tested for prediction accuracy in an independent BM verification sample set (N = 24) from ECOG AML trials to select the final classifier, which was a significant predictor of CR/CRi (area under the receiver operating characteristic curve AUROC = 0.76, p = 0.01). The selected classifier was then validated in the SWOG BM Validation Set (AUROC = 0.72, p = 0.02). Importantly, a classifier developed using only clinical and molecular inputs from the same sample set (DXCLINICAL2) lacked prediction accuracy: AUROC = 0.61 (p = 0.18) in the BM Verification Set and 0.53 (p = 0.38) in the BM Validation Set. Notably, the DXSCNP classifier was still significant in predicting response in the BM Validation Analysis Set after controlling for DXCLINICAL2 (p = 0.03), showing that DXSCNP provides information that is independent from that provided by currently used prognostic markers. Taken together, these data show that the proteomic classifier may provide prognostic information relevant to treatment planning beyond genetic mutations and traditional prognostic factors in elderly AML.

Show MeSH