Variable Pathlength Fiber-Optic Spectrophotometry for Protein Determination in Immunoglobulin Concentrates

Peer-reviewed Article submitted: July 13, 2015  Article accepted: Aug. 17, 2015

AbstractAs an innovative and time- and resource-saving single-step technique that avoids sample pre-dilution, the use of hazardous reagents, and high-power consumption, SoloVPE fiber-optics spectrophotometry may be able to substitute or replace the current Kjeldahl protein determination procedure for 10% IgG in the future.

ALL IMAGES/FIGURES ARE COURTESY OF THE AUTHORS.

For decades, spectrophotometry has been used to accurately and precisely measure purified protein concentrations, with the extinction coefficient (usually at 280 nm) determined either experimentally or calculated from amino-acid composition. To reduce the overall instrumental uncertainty in absorbance measurement, samples require dilution to an absorbance of 0.6–0.7/cm. The uncertainty in the sample preparation step can be minimized in gravimetric dilution, but measurement of a larger series of samples generally involves considerable work to ensure flasks and cuvettes are clean.   To overcome the limitations of conventional cuvette-based spectrophotometry, a spectrophotometric technique based on measurement of the incremental absorbance upon path length variation has recently been introduced (1). This variation is achieved by moving an optical light-guide fiber immersed into the sample solutions in a stepwise fashion, with the absorbance readout at every step and the calculation of the slope from the absorbance vs. distance plot. This technique is called “slope spectroscopy” by the manufacturer and is intended for use in measuring concentrated sample (protein) solutions without dilution. The SoloVPE system (C Technologies, Inc.) has been commercially available for a few years and has been evaluated mainly for concentration measurement of recombinant proteins such as monoclonal antibodies (2).
  In plasma fractionation, an interesting candidate for assessment of the slope spectroscopy technique would be immunoglobulin G for intravenous administration (IVIG), highly purified by chromatography and formulated as a liquid concentrate. Gammagard liquid is a polyvalent IgG product, containing 10% protein of more than 98% purity in a 250 mM glycine solution, pH=4.8 (3). Currently, the protein is measured by differential Kjeldahl determination of the total and the non-protein nitrogen content (4). Despite the automation in reagent dosing and titration, which minimizes handling of hazardous substances such as sulfuric acid, sodium hydroxide solution, and hydrochloric acid, the Kjeldahl technique remains a time- and energy-intensive method and requires effective removal of acid fumes.
  While not specifically in use for Gammagard liquid, the other available nitrogen-based technique, the Dumas method, does not require hazardous liquids, but instead an oxygen-helium stream to support sample combustion in steel crucibles or single-use tin capsules, which are heated electrically in a furnace to approximately 1000 °C (5). Although the Dumas method, with its higher throughput, has been almost fully automated, the power requirement and reagent consumption remain significant.
  The batch-to-batch consistency of the protein composition allows total protein measurement by ultraviolet (UV) absorbance as an alternative to nitrogen-based techniques in a sample dilution containing about 0.5 mg IgG/mL with physiological saline as the diluent and blank. 10 percent IgG must be diluted approximately 200-fold in a volumetric flask. The turbidity is corrected by measurement at 320 nm, and the extinction coefficient E

_0.1%

is calibrated to 1.399/cm on the Kjeldahl data.
  For Kjeldahl digestion, Dumas combustion of the protein solution, and UV absorbance measurement with gravimetric dilution, the sample density needs to be first measured in an oscillating tube densimeter. If the slope-spectroscopy technique (by way of the of SoloVPE device) yields equivalent results, time-consuming sample preparation and measurement steps can be reduced to a single-step procedure, cleaning of the equipment may be eliminated, consumption of reagents can be decreased, and the electrical power obviated.
  Equivalence of this assay technique with the Kjeldahl method must be demonstrated first by adherence to the AOAC acceptance criteria for precision and accuracy: At an analyte concentration of 10%, recovery must remain within 98–102% and precision no higher than 2.8% (6, 7); accuracy is confirmed by agreement with an independent, validated reference method (such as the Kjeldahl digestion). Inter- and intra-assay precision must be determined by six-fold repeated measurements as outlined in the respective regulatory guidelines; linearity should be at least at 80–120% of the working level. Furthermore, due to the commercial importance of IVIG, the bias (difference in results) between both methods should be as small as possible, as shown by the Bland-Altman statistical analysis of the datasets (8).
  A 20% subcutaneous IgG concentrate, formulated in the same slightly acidic environment as the corresponding licensed 10% IVIG concentrate, is now on the market, thus, 5–25% IgG would cover the entire range during production. For rheological investigation, even higher IgG concentrations (up to 38%) have been obtained experimentally (9).
   

Materials and methods

Thirty-eight batches of 10% IVIG (Gammagard liquid), with known protein concentration as determined by Kjeldahl method with solution density measurement, were obtained in the original bottles, from which aliquots were drawn antiseptically with a syringe. An experimental ultra-high IgG concentrate (target: more than 30%) was prepared by centrifugal ultrafiltration (Centriprep 10 kDa) of an 0.03% azide-preserved aliquot, and the permeate retained to mix dilutions (5%, 10%, 15%, 20%, and 25% IgG) with the stock concentrate (32%). Of these preparations, protein concentrations were determined by UV absorbance measurement at 280 nm (with turbidity correction at 320 nm) and by Dumas combustion method in a VarioMax analyzer (Elementar), with the permeate (glycine/azide) as the base value. Solid acetanilide served as the standard for the nitrogen content, which was assumed to be 16% for Kjeldahl and Dumas determination (10).   For conventional UV-absorbance measurement, a Hitachi U-3000 spectrophotometer with standard 1-cm quartz cuvettes and a Hitachi U-3310 high-sensitivity spectrophotometer with 0.1-mm thin-layer quartz cuvettes in a spring-clamp holder were used.
  The SoloVPE fiber-optic spectrophotometer, comprising the variable pathlength fiber/detector unit and an Agilent Cary 60 as the light source/spectrometer unit, was provided by C Technologies, Inc., together with the single-use plastic well cuvettes and the polyimide-coated 0.6-mm diameter quartz “fibrettes” as the consumables.  

Figure 1: Photograph of a light path from the SoloVPE fibrette dipping into fluorescein solution.

Variable pathlength vs. conventional spectrophotometry
In the SoloVPE device, monochromatic light fed from the fiber-optic light guide passes through the quartz fibrette and is absorbed by the sample solution in the adjustable path between the fibrette end window and the plastic well cuvette (

Figure 1

). A silicon photodiode at the bottom of the cuvette cavity measures the transmitted light intensity. In the solution, the light pathlength

d

can be increased in increments as small as 0.05 µm, and the absorbance values

A

are read out at all 10 measurement points. To remain within a linear regression (R

²

≥0.999), at least five results are required for slope calculation. The slope ∆A/∆d obtained from the absorbance vs. pathlength regression calculation is then divided by the mass specific or molar extinction coefficient (E or ε) to calculate the mass or molar concentration, respectively.   Spectrophotometric measurement is inherently limited by both dark signal and noise of the detector as well as by the bit-depth of the analog/digital signal converter. A deviation from linearity caused by monochromator stray light may be readily observed above 3 absorbance units (i.e., 0.001 residual transmission). High-sensitivity spectrophotometers such as the Hitachi U-3310 with its additional auxiliary stray light-eliminating monochromator may extend the upper absorbance range limit from about three to four absorbance units (i.e., from 0.001 to 0.0001 residual transmission), above which the thermal detector noise of the common red-sensitive photomultiplier (PMT) becomes dominant. For routine ultraviolet-visible spectrophotometry (UV-VIS) instruments, equipping the detector with a blue-sensitive PMT and cooling it from ambient temperature to -20 °C for example, is not a viable option.
  A straightforward and presumably user-friendly approach to overcome these instrumental limitations was proposed through pathlength reduction by the use of thin-layer cuvettes. Several manufacturers offer cuvettes with a pathlength of 0.5, 0.2, and even 0.1 mm, and a detachable cover window. The suitability of such a technique for measuring a monoclonal antibody up to a concentration of 90 mg/mL has been described by Watson and Veeraragavan (11). Although an absorbance value of 1.4 for 10% IgG in a 0.1-mm cuvette would remain unaffected by spectrophotometer interference from stray light, dark signal, or noise, the authors’ experience discourages from routine use of such cuvettes because of the delicate handling and cumbersome cleaning involved, and the practical impossibility of applying sample concentrations above 20% IgG due to their high viscosity.
 

Results and discussion

10% IgG solutions and inter-assay precision

Table I: Comparison of SoloVPE and Kjeldahl protein concentration results for 10% IgG batches with recalculation of the extinction coefficient E as slope/Kjeldahl protein.

For all 38 batches of 10% IgG, protein concentration data obtained using the SoloVPE and the Kjeldahl method were in good agreement (see

Table I

). The recalculated average extinction coefficient of E

_0.1%

as 1.404 ± 0.013/cm corresponds well to the Kjeldahl-calibrated value of E

_0.1%

=1.399/cm for the total protein (TP) UV measurement in 200-fold diluted solution, especially when considering the inherent measurement uncertainty.
  A statistical Bland-Altman comparison between SoloVPE and Kjeldahl protein concentration data was performed. The results indicate that both methods agree well (at a 95% confidence interval). The mean bias was -0.33% with 95% limits of agreement ranging from 1.56% to 2.21%, both of which do not exceed the acceptable recovery limits of 98% and 102% for an analyte concentration of 10%.
  The intra-assay precision test for 10% IgG solutions, run in a separate series, consisted of six consecutive readouts of the same sample well cuvette with the same fibrette. As the sample solution can be used directly without any preparation procedure, the coefficient of variation did not exceed 0.5% (see

Table II

).
 

Experimental 5-25% IgG solutions and intra-assay precision

The dilutions were prepared by weighing from the 32% stock concentrate and the permeate, based on the initial protein concentrations and density values. For comparison, the dilutions were analyzed by the Dumas method in a VarioMax combustion automat with density-based sample dosage. The stock TP UV concentration was determined using the single TP UV measurement as 325.46 mg/mL, and averaged from all six TP UV measurements (five dilutions and the stock), based on the respective overall dilution factor, as 321.88 ± 3.59 mg/mL. For the SoloVPE measurements without dilution, this particularly viscous stock was difficult to fill into the sample well cuvettes without producing bubbles.
 

CLICK IMAGE TO ENLARGE Table II: Intra-assay precision results for six 10% IgG batches (lot numbers the same as in Table I). RSD=relative standard deviation.

While the 5% and 10% solutions gave an excellent slope regression for the SoloVPE measurement with all 10 data points on the regression curve, the higher concentrations required-with increasing concentration-moderate to considerable “slope tweaking” by omission of visually deviant data points, usually at absorbance values >1.6 and at times also of the lowest value at the “foot” (see

Figure 2

).
 

Figure 2: Manual "slope tweaking" by omission of absorbance values deviant from linearity (A>~1.7 AU) may improve accuracy for high-protein concentrations.

Figure 2: Manual "slope tweaking" by omission of absorbance values deviant from linearity (A>~1.7 AU) may improve accuracy for high-protein concentrations.

Only with this recalculation of the slope can the difference between the SoloVPE and the other methods be reduced for the highest concentrations, to approximately -7%. Intra-assay precision, determined for these samples by measurement on six days, was reasonable, with relative standard deviations (RSDs) below 1.5% (see Table III). Table III also shows the extinction coefficient E_0.1%, recalculated using these data.

CLICK IMAGE TO ENLARGE Table III: Protein concentrations as measured by Dumas, total protein ultraviolet (TP UV), and SoloVPE method, and SoloVPE inter-assay precision results for six experimental 5-32% IgG solutions. RSD=relative standard deviation.

Interestingly, the variation in the recalculated extinction coefficient E_0.1% (based on the TP UV concentrations) suggests a concentration-dependence with a possible optical effect of macromolecular crowding, in particular a reduced increase in scatter (and turbidity) at extremely high concentrations (12, 13). A corresponding shift in light attenuation (i.e., the sum of absorbance and turbidity) has to our knowledge not been described so far.

While mathematical modeling and experimental evidence of such effects for IgG are beyond the scope of the present study, the data obtained indicate that the extinction coefficient for the TP UV method of 1.399/cm, as calibrated for 10% IgG in 250 mM glycine, cannot be applied to a corresponding 20% IgG concentrate and its even higher-concentrated ultrafiltration stock using the SoloVPE technique.  

Instrument performance and system suitability tests

Wavelength accuracy was determined daily using the sealed cuvette containing the Ho(ClO

₄

)

₃

in perchloric acid in the standard cuvette holder, for which the deflection mirror had to be shifted and the connector cable changed at the start of every measurement day. Alternatively, a separate SoloVCA adapter (C Technologies, Inc.) can be connected via the fiber-optic light guide to the spectrophotometer. From days four to five (during the weekend), the SoloVPE spectrophotometer was turned off (as the Cary 60 does not include a high-voltage supplied detector with a photomultiplier tube, it may be left on overnight during the week). The wavelength position was read out manually from the software’s window with the tracer cursor. At the instrument’s fixed slit width of 1.5 nm, the wavelength error was - 0.5 to -0.8 nm for the 278.11 nm and -0.7 to -1.1 nm for the 287.40 nm absorption maximum (see

Table IV

) (14).
 

CLICK IMAGE TO ENLARGE Table IV: Wavelength results determined using the Cary 60 spectrophotometer and photometric check results from the SoloVPE; n.d.=no data.

To establish a photometric system suitability test, the absorbance A of a water-soluble phenolic compound solution was measured at 280 nm in 0.1 mm cuvettes as 117.25/cm with the Hitachi U-3310 spectrophotometer and as 114.00/cm with the Hitachi U 3000 spectrophotometer, and in the 0.2 mm cuvette as 113.00/cm in the U-3310. In the SoloVPE, the range was consistently between 103 and 106/cm (see

Table IV

); the reason for this difference is unknown.
 

Conclusion

A reasonable agreement between the SoloVPE and Kjeldahl method was obtained when measuring IgG concentrations with the current release methods at adequate precision. Although the extinction appears to be influenced by macromolecular crowding effects depending on concentration, the extinction coefficient E

_0.1%

of 1.399/cm for 1 mg IgG/mL, as calibrated for the TP UV method, remains valid using the SoloVPE technique at a concentration of 10% to fulfill the AOAC criteria for recovery and precision. As recalibration of the extinction coefficient is not required, variable pathlength spectrophotometry appears to be a viable instant total protein method to determine the current 10% IgG concentrate. For higher concentrations such as a 20% IgG solution, however, specific recalibration of the extinction coefficient is essential.
  As evident from the raw measurement data for high IgG concentrations, the variable pathlength technique might benefit from further instrumentation improvements, such as a higher incident light beam intensity or lower detector dark signal, to overcome the limited absorbance range. Before implementation of such modifications into a new instrument generation, however, the user-customizable slope calculation algorithm should be set so as to omit absorbance values >1.7/cm as observed with the current version.
  As an innovative and both time- and resource-saving single-step technique that obviates sample pre-dilution, the use of hazardous reagents, and high power consumption, SoloVPE fiber-optic spectrophotometry may be able to replace the current Kjeldahl and Dumas protein determination procedures for other protein species, with a constant composition and extinction coefficient.
 

Acknowledgements

The authors thank Joe Ferraiolo and Larry Russo (C Technologies, Inc.) for SoloVPE instrument setup, user training, and technical support; Laurence Plapied, Grégoire Bouchez, and Christophe Carnewal (Baxalta Lessines Technical Operations) for providing the Gammagard liquid samples and the Kjeldahl protein concentration data; and Brigitte Neubauer-Watzal, Daniel Bergmann, and Robert Ras (Baxalta Vienna Quality Control Department) for the Dumas analysis. Karima Benamara is acknowledged for editing the manuscript.

References:
1. M.C. Salerno, T. Shih, and C. Harriston/C Technologies, Inc., “Interactive Variable Pathlength Device”, US patent application US2009/0027678, January 2009.
2. S. Huffman, K. Soni K, and J. Ferraiolo, Bioproc. Int. 12 (8), pp. 66–73 (2014).
3. W. Teschner et al., Vox Sang. 92, pp. 42–55 (2006).
4. J. Kjeldahl, Fresenius' Z. Anal. Chem. 22, pp. 366–382 (1882).
5. P. Blondel and L. Vian, Ann. Pharm. Fr. 51 (6), pp. 292–298 (1993).
6. L. Huber, LC-GC Int. 1998 (2), pp. 96–105 (1998).
7. I. Taverniers, M. de Loose, and E. van Bockstaele, Trends Anal. Chem. 23, pp. 536–551 (2004).
8. J.M. Bland and D.G. Altman, Lancet 8476 (1), pp. 307–310 (1986 Feb. 8).
9. V. Burckbuchler et al., Eur. J. Pharm. Biopharm. 76, pp. 351–356 (2010).
10. E. Brand, B. Kassell, and L.J. Saidel, J. Clin. Invest. 23, pp. 437–444 (1944).
11. L. Watson and K. Veeraragavan, Biopharm Int. 27 (2), pp. 26–37 passim (2014).
12. C. Fernandez and A.P. Minton, Anal. Biochem. 381 (2), pp. 254–257 (2008).
13. C. Fernandez and A.P. Minton, Biophys. J. 96, pp. 1992–1998 (2009).
14. J.C. Travis et al., J. Phys. Chem. Ref. Data 34, pp. 41–56 (2005).

ALL FIGURES ARE COURTESY OF THE AUTHORS.

About the Authors
Heinz Anderle is senior scientist and Alfred Weber is senior manager; both at Baxalta Innovations GmbH.

Article DetailsBioPharm International
Vol. 29, No. 12
Pages: 42–50
Citation: When referring to this article, please cite it as H. Anderle and A. Weber, "Variable Pathlength Fiber-Optic Spectrophotometry for Protein Determination in Immunoglobulin Concentrates," BioPharm International29 (1) 2016.