AI-64310-CRFT

Figure 11 summarizes the qualitative aspects of the

quantitation as a function of the levels spiked and the

confidence in the isotopic overlap between experimental

and theoretical values. The response at each level

represents over 30 measurements where all charge states

across each quantitation curve were considered. As

expected, the lower levels resulted in a lower average dot

product correlation coefficient (0.58 for all insulin

variants spiked at 1.5 pM). However with six isotopes

considered, the coefficient was discriminatory and

accurately defined potential background interference that

would disrupt the true isotopic distribution in the four

most abundant isotopes. Even at very low analyte levels,

the quality of the quantitation is quite high. All other

levels generated average dot product correlation

coefficients greater than 0.8. The average dot product

coefficients for porcine measurements were consistently

above 0.97 for the 50 pM level.

Quantitation curves were generated based on user-defined

levels. To best match the relative AUC values, each charge

state was used to normalize the corresponding charge

state for human insulin. This provided a significant benefit

as relative abundance values are generally not equivalent,

and simply summing and normalizing AUC response per

target will bias towards the most abundant charge state.

Normalizing at each charge state and showing the

summed response for each individual charge state, takes

advantage of HRAM data to increase the discriminatory

power of the method.

As shown in Figure 9, the top right table shows the

cumulative quantitation results for the row highlighted

and contains all of the values used to evaluate the

quantitation, including the %RSD and calculated

amounts. Specific values can be added or subtracted.

When a row in the table is selected (highlighted in blue),

the accompanying results are graphically displayed in the

bottom half of the window. The bottom left of Figure 9

shows the normalized quantitation curve for human

insulin. The curve is weighted by 1/x and the equation is

displayed. The bar chart (bottom middle) displays the

relative AUC values for the six isotopes across the spiked

levels. As reported in the upper right table, only the lowest

spiked amount (1.5 pM) could not be used to identify all

six isotopes. The graph on the lower right displays the

XIC data for any single RAW file selected and can be

changed by clicking on any point in the quantitation curve

or bar chart.

The cumulative results from individual quantitative curves

for the first sample set were overlaid to demonstrate that

the workflow was global in its ability to quantify the

different insulin variants (Figure 10). Each curve, when

normalized to porcine insulin, had a linear regression of

0.98 or greater for all precursor charge states, isotopes,

and reported summed AUC values. The difference in

slopes was attributed to the relative ionization efficiency

and antibody binding coefficients of the different insulin

analogs. The workbook used to process the insulin

quantitation curves spiked into PBS/BSA was also able

to process the insulin quantitation curves spiked into

human plasma.

200

400

600

800

1000

1200

AUC Ratio [Insulin Variant:Porcine]

Spiked Amount of Insluin Analog (pM)

Humulin S

Lantus

Apirda

Bovine

Novorapid

0.2

0.4

0.6

0.8

1.2

1.4

1.6

Figure 10. Comparative quantitation curves for all insulin analogs spiked into PBS/BSA

matrix. The AUC values were normalized against the measured AUC value for porcine.