Blood-based proteomics for personalized medicine: examples from neurodegenerative disease

Figures & data

Figure 1. 2D gel of human blood serum.

The dynamic range and reproducibility of quantitative 2D gel electrophoresis of human blood serum. More than 1500 protein spots are resolved and quantitated within the 2D gel pattern of human blood serum. Typical quantities, ranging from <100 to >15,000 ppm, are indicated to demonstrate the dynamic range (black arrows).

Figure 2. Differential diagnostics through multivariate biostatistics of blood serum concentrations of groups of proteins using quantitative 2D gel electrophoresis.

Disease-specific discrimination was accomplished using both canonical discriminant analysis (top) and linear discriminant analysis (bottom), showing discrimination between patients with AD, PD and ALS (A) and patients with AD, AD-like and normal controls (B).

AD: Alzheimer’s disease; ALS: Amyotrophic lateral sclerosis; PD: Parkinson’s disease. Data from [14,15].

Figure 3. Statistical pattern of N3 serum protein biomarker in different neurodegenerative diseases.

(A) Box and whiskers quantitative level and (B) capability of differentiation between AD and AMC.

AD: Alzheimer’s disease; AMC: Age-matched controls; PD: Parkinson’s disease; ROC: Receiver operator characteristics.

Figure 4. Statistical pattern of N7 serum protein biomarker in different neurodegenerative diseases.

(A) Box and whiskers quantitative level and (B) capability of differentiation between AD and PD.

AD: Alzheimer’s disease; ADL: AD-like and mixed; AMC: Age-matched controls; PD: Parkinson’s disease.

Appendix 1. Reproducible quantitation of 13 different protein spots.

Biomarker	Number	Mean ± standard deviation error	Coefficient of variation ≤20%
M1	14	13542	711
M2	14	3853	140
M3	14	1413	52
M4	14	1015	49
M5	14	678	28
M6	14	655	33
M7	14	595	31
M8	14	469	26
M9	14	359	16
M10	14	209	11
M11	14	129	5
M12	14	106	6
M13	14	72	4

Illustrates the reproducibility of quantitation of protein spots over the dynamic range of the 2D gel assay of human serum depicted in Figure 1. Reproducible quantitation of 13 different protein spots covering the dynamic range was obtained by running the same sample of human blood serum on 14 different 2D gels. The gels were run by different technicians on different days over 2.5 months (n = 14; range 72–13,542 ppm). The reproducibility was demonstrated with a coefficient of variation of ≤20% (n = 14), independent of the protein concentration, where the lowest concentration spot, 72 ppm, is approximately tenfold higher than the limit of detection (100 pg/spot = ∼5–10 ppm) of the assay.

Appendix 2. Mean ± standard error of N3 in different neurodegenerative diseases and statistical comparison to age-matched control.

N3	Number	Mean (ppm)	± standard error	AMC (%)	ANOVA-P
AMC	75	1047.7	36.88	100	< 0.0001
Alzheimer’s disease	115	656.0	27.01	63
Parkinson’s disease	12	230.8	42.58	22
Alzheimer’s disease-like plus mixed	12	355.4	48.89	34

AMC: Age-matched control; ANOVA-P: p-value of analysis of variance.

Appendix 3. Receiver operator characteristics AUC and cut-off value that provides maximum sensitivity and specificity for N3 when used as a single biomarker to differentiate between Alzheimer’s disease and age-matched control.

N3 (Alzheimer’s disease < AMC)	ROC (N3 < cut-off)	Sensitivity	Specificity	Area	SE	ROC-P
AMC			64.0%	0.71	0.022	<0.0001
Alzheimer’s disease	804	64.1%

See (Figure 3).

AMC: Age-matched control; ROC: Receiver operator characteristics; ROC-P: p-value of ROC; SE: Standard error.

Appendix 4. Mean (ppm) ± standard error of N7 in different neurodegenerative diseases and statistical comparison to age-matched control.

N7	Number	Mean (ppm)	± standard error	AMC (%)	ANOVA-P
AMC	75	1675	52.20	100
AD	115	2000	48.48	119	<0.0001 AD > AMC
PD	12	1439	119.85	86	<0.0001 AD > PD
AD-like plus mixed	12	1624	138.90	97	<0.0001 AD > ADL

AD: Alzheimer’s disease; AMC: Age-matched controls; ANOVA-P: p-value of analysis of variance; PD: Parkinson’s disease.

Appendix 5. Receiver operator characteristics AUC and cut-off values that provide maximum sensitivity and specificity for N7 when used as a single biomarker to differentiate AD from PD, AD-like or age-matched control.*

N7 (AD < AMC)	ROC (N7 < cut-off)	Sensitivity (%)	Specificity (%)	Area	Standard error	ROC-P
AD > AMC	1698	56.2	56.4	0.61	0.024	<0.0001
AD > PD	1500	71.9	72.2	0.72	0.0244	<0.0001
AD > ADL	1693	56.8	58.3	0.62	0.049	<0.007

*Best performance for N7 when used to differentiate between AD and PD.

See (Figure 4).

AD: Alzheimer’s disease; ADL: Alzheimer’s disease-like; AMC: Age-matched controls; PD: Parkinson’s disease; ROC-P: p-value of receiver operator characteristics; ROC-P: p-value of ROC.

Appendix 6. Mechanism of neuronal degeneration measured in blood.

Mechanisms	AD-like mixed dementias	Parkinson’s disease	Alzheimer’s disease
N1: Autoimmune activity	Unchanged	Increased	Increased
N2: Innate inflammatory activity	Increased	Increased	Increased
N3: Inhibition of extracellular signals for apoptosis	Decreased	Decreased	Decreased
N4: Inhibition of escape from cell degeneration	Increased (stroke-related and mixed dementias only)	Increased	Increased
N5: Innate inflammatory activity	Increased (stroke-related and mixed dementias only)	Increased	Increased

Appendix 7. Disease processes measured in blood of ALS patients.*

Mechanism of neuronal degeneration	Amyotrophic lateral sclerosis
	Familial	Sporadic
Neuronal oxidative stress and apoptosis	Intracellular primary	Inhibited pathway
Autoimmune, innate and/or acquired inflammation	Secondary innate and acquired	Primary autoimmune

*See [13,14].