- Email: [email protected]
Molecular Profiling and Weight Determination of Heparins and Depolymerized Heparins
Molecular Profiling and Weight Determination of Heparins and Depolymerized Heparins A. AHSAN*~, W. JESKE*, D. HOPPENSTEADT*, J. C. LORMEA~, H. WOLF'§, AND J. FAREED" Received October 12, 1994, from the *Departments of Pathology and Pharmacology, Loyola University Medical Center, 2160 S. First Avenue, Maywood, IL 60153, Ganofi Recherche, Toulouse, France. Accepted for publication January 31, 1995@. 5 Permanent address: Sandoz A. G., Nurenberg, Germany. Abstract 0 The recently proposed calibrant LHN-1 (lot F537; henceforth designated F537), for the molecular weight (MW) determination by highperformance size-exclusion chromatography of heparins, is shown here to have a range too narrow to allow for the accurate MW determination of all low molecular weight heparins (LMWHs). We have recently demonstrated, by this same methodology, that a chemically degraded benzyl ester of unfractionated heparin, heparin mass calibrator (HMC), is a better calibrant. Weight-average MW, number-average MW, peak MW, and dispersity values were calculated with F537, HMC, and by a reference narrow-range-calibration method for various LMWHs and unfractionated heparins. Values for these parameters determined with HMC were not significantly different from those determined by the reference method until the MW of the substance exceeded 15.0 kDa. In contrast, the MW profile obtained with F537 was appreciably different from that obtained by the reference method for samples with MWs >7.5 kDa. The range exhibited by HMC should allow this calibrant to be used for both LMWHs and unfractionated heparins.
Heparins are polymeric agents consisting of sulfated polysaccharide chains with broad molecular size dispersi0n.l The molecular weight (MW) of these chains ranges from 3 to 25 Low molecular weight heparins (LMWHs), formed either by chemical or enzymatic depolymerization of heparins, have MWs ranging from 3 to 8 kDa.l Molecular size is important in the determination of biological activity of any particular h e ~ a r i n ,so ~ various .~ methods have been developed to estimate this property. These techniques include measurement of osmotic pressure, sedimentation-diffusion, viscosity, and light-scattering behaviors.* Among these methods are the gel filtration (used here) and the viscometric procedures, which, however, must be calibrated by ultracentrifugal and light scattering method^.^ The determination of the MW of heparin preparations by high-performance size-exclusion chromatography (HPSEC) requires that the column(s) employed be properly calibrated with like molecules. In our laboratory, 19 oligosaccharide fractions isolated from native heparin are used to generate a log MW versus retention time calibration curve. These fractions are narrow MW distribution standards with dispersities of The oligosaccharide fractions cover a MW range of 1.3to 22.5 making them applicable for the MW profile determination of LMWHs and unfractionated heparins (UFHs). The MWs of these calibrators were determined by low-angle laser light scatter (LALLS) among other methods. Generation of such well-characterized polymer standards requires extensive laboratory time and is the main disadvantage of HPSEC methods6 A new procedure for obtaining the calibration curve for the HPSEC determination of the MW profile of LMWHs has been proposed by van Dedem and N i e l ~ e n .In ~ this procedure, a single polycomponent mixture (F913B) of heparin oligosaccharide chains, produced by heparinase-catalyzed partial < l . l . 5 3 6
@
Abstract published in Advance ACS Abstracts, April 1, 1995.
724 /Journal of Pharmaceutical Sciences Vol. 84, No. 6, June 1995
degradation of UFH, is used to determine calibration values. F913B is a product of partial heparinase digestion of UFHs. Because of the specificity of heparinase, the elution profile of F913B contains a series of regularly spaced peaks in the lower MW regions. Each of these peaks differs from its adjacent peak by one disaccharide unit, which has an assumed MW of 600 kDa. We have recently shown that this elution profile gives estimates of the MWs of various LMWHs that are in agreement with those determined by the NRC method, provided that the components of the LMWH mixture are <7.5 kDa. This procedure may be acceptable for measuring LMW components in the LMWHs [however, another calibration method would be required to measure the medium, (MMW) and high molecular weight (HMW) entities] as well as for obtaining MW profiles of UFHs. The percentages of HMW components determined by this method were consistently overestimated. Several other calibrators have since been developed for use with the technique of van Dedem and N i e l ~ e n .Ahsan ~ et d.l0 recently reported another calibrator produced by partial chemical degradation of the benzyl ester of UFH. The elution profile of this calibrator is considerably different from that of F913B obtained by the heparinase degradation method.1° The new calibrator, tentatively named HMC, gives more accurate MW parameters and dispersities. A second heparinasedegraded calibrator, F537, has also been proposed for determining the MW profile of heparins.ll A comparison of the elution profiles of F913B and F537 shows that the latter exhibits somewhat larger absorbances a t lower retention times and, therefore, contains more of the HMW material. Presumably, F537 was made by letting heparinase degradation run for a shorter time than in the case of F913B. In this paper, we compare the MW and dispersity data obtained for LMWHs and UFHs with the two newer calibrators, HMC and F537, with the data obtained by the reference narrow-rangecalibration (NRC) method used in our laboratories.
Experimental Section Calibration curves were prepared by running the appropriate compounds on a Waters 845 GPC-HPLC system (Millipore-Waters, Lexington, MA) equipped with Expert Ease software designed for polymer analysis. Each calibrator was dissolved in 0.5 M NaZS04 to give a lO-mg/mL solution. The HPLC system consisted of a VAX 3100 computer (Digital Corp.), a L A C E interface module, two 510 HPLC pumps, a 712 WISP autoinjector, a R401 differential refractometer, and a 484 tunable absorbance detector (all from Waters). The UV and the refractive index (RI) detectors were linked in series, with the outlet end of the columns attached to the UV detector. A 2O-pL aliquot of each sample was injected onto a joint column, TSK G3000SW and TSK G2000SW (Tosoh Corp., Japan), connected in that order. The protocol used for the runs with the F537 sample was the same as previously described,l' except that the aforementioned TSK columns were used instead of the Waters PAC I 60 and I 125 columns used in the other study. The method of acquiring the calibration curve data was very similar t o the one described in a previous paper.7 Both procedures used the TSK columns just mentioned, the differences being in the concentration of the mobile phase (0.5 M in our study and 0.2 M in the vanDedem and Nielsen
0022-3549/95/3184-0724$09.aO/O
0 1995, American Chemical Society and American Pharmaceutical Association
study7)and in the injection volume (20 pL in our study and 25 pL in the vanDedem and Nielsen study7). The flow rate for the mobile phase (0.5 M NazS04) was 0.5 m u m i n and the run time was 65 min. The UV determinations were made at 234 nm, and all analyses were done a t room temperature. This protocol was also followed for runs with HMC. For the partially degraded calibrators HMC and F537, MW assignments for each peak were made according to the method of van Dedem and N i e l ~ e n .Briefly, ~ the strengths of the RI and UV signals of each peak in the calibration mixture were determined. The ratio of these two values was then related to MW based on the assumption that the second-to-the-last peak of each mixture is that of a tetrasaccharide with a MW of 1.2 kDa, or the third one is a hexasaccharide (1.8 m a ) , and so on until each peak was assigned a MW. The criterion for selection of a particular peak as a reference point is that it agrees best with the rest of the peak ratios. The F537-derived calibration curves were prepared assuming the fourth peak from the low molecular end to be that of the octasaccharide. This assumption was in accord with that of Nielsen.l1 For the 19-standard-NRC method, the UV detector was set at 205 nm, where it is most sensitive, and the log MW of each calibrant was plotted against the retention time. A third-order regression equation was fitted to the data points with the Expert Ease software. Analyses of heparins and LMWHs used in this study were performed in the same way as for the calibrators. Following each run, the elution profile of each sample was analyzed by the three calibration methods. In addition to calculating the weight-average MW of each sample, the software was used to determine several other MW parameters based on the elution profile, including numberaverage MW, peak MW, and dispersity. The heparins and LMWHs used in this study are all sodium salts, supplied as white powders. The majority of these substances are marketed as injectables by pharmaceutical firms listed in the tables. For statistical analysis, MW data was separated into two groups ('8.0 kDa and '8.0 kDa). Data obtained for the various parameters by the three calibration methods were compared by independent analysis of variance (ANOVA). For each MW group,^ values of x0.05 were considered statistically significant (Primer of Biostutistics. Computer Program; McGraw-Hill: San Diego, CA). However, as HPSEC is not a n absolute method of MW determinations, the uncertainty expected in typical determinations of MW parameters will also vary with the method used to obtain the MW of the reference calibration standards. For instance, with LALLS, a popular absolute method of MW determination, a 10%uncertainty may be e ~ p e c t e d . ~
Figure 1-The comparative UV elution profiles of the (A) HMC and (B) F537 calibrator. The x-axis represents the elution time in min, the y-axis denotes the detector response in percentage (%), and the maximum signal in each case is given the value 100.
5.0
I
NRC v HMC V F537
I
20' 25
30
35
40
45
50
Retention time (min)
Results
Figure 2-Calibration curves generated by (0)the NRC method, (v)the F537 calibration method, and (v) the HMC method. Each of the calibration curves has a correlation coefficient of 9.999. The x-axis denotes the elution time in min, and the y-axis denotes the log MW.
Like F913B, F537 is prepared by depolymerization with heparina~e.~a~Jl The enzymatic action partially cleaves 1,4bonds between sulfated glucosamine and iduronic acid residues, the latter producing chains containing 4,Fi-unsaturated uronic acid groups at the nonreducing end, which absorb strongly a t 234 nm in the UV region. Depending on the conditions, a greater or smaller fraction of the bonds will be split. The fragments produced are multiples of disaccharide units, a consequence of the alternating uronic acid-glucosamine structure of heparin. A comparison of the HPSEC elution profiles of F537 and F913B shows that the former has somewhat more of the longer polymer chains than F913B. HMC is produced by benzylation and alkaline hydrolysis of UFH1s8 obtained from porcine intestinal mucosa. This method of degradation also leads to the formation of 4,5unsaturated uronic acid end groups, which absorb strongly a t 234 nm. The elution profile of this calibrator at 234 nm is considerably different from that obtained by heparinase degradation (Figure 1).Much larger absorbances are seen for HMC prior to an elution time of 35 min relative to the absorbances in the same region with the elution profile of F537. This result indicates that there is a larger amount of material a t the HMW ends of the profiles for HMC than for F537. The NRC calibrators are described in a previous publication.10 Briefly, they were made by fractionation of heparin
on Ultrogel AcA 44 agarose acrylamide (LKB-ProducteurAB) gel-permeation chromatographic c01umns.~This method is based on the procedure of Johnson and Mulloy.lZ Each of the individual narrow peaks in Figure 1 gives its retention time on the x-axis and the logarithm of the MW on the y-axis, giving the calibration curve. The best-fit calibration curves produced with each calibrator, the retention times of the individual narrow range standard runs giving the NRC curve, and the regularly spaced peaks of the F537 and HMC runs giving their respective curves are shown in Figure 2. All the curves had a correlation coefficient of >0.999. The curves are not identical, however. The HMC and NRC calibration curves are nearly identical at retention times between 36 and 45 min. This retention time corresponds to a MW between 2.5 and 8 kDa, the region where most LMWH components elute. The calibration curve of F537 is somewhat different. Computer software application of the three calibration curves so prepared to the elution profiles of the samples listed in the tables gave the molecular parameters. Table 1contains the MW profile of several compounds with a weight-average MW of 4 . 0 kDa. Statistically significant differences were observed only in the dispersity values, with those calculated by the F537 method being higher than those obtained by the NRC method 0, = 0.045) or the HMC method (p = 0.011). In the very LMW region (i.e., the first three entries of Table l), there is agreement between the F537-derived values and Journal of Pharmaceutical Sciences / 725 Vd. 84, No. 6,June 1995
Table 1-Comparative Analyses of LMWHs by Narrow Range Calibrator, F537, and Heparin Mass Calibrator Methods sample name and code CY 231 PQ2ZH AprosulateC CY 222b PIlWH Enoxaparind Fraxiparinb Flagmine Abbott' Std., 6200 Heparing Lot E, Upjohn Novoh 6076-4
Molecular Weight (kDa)
calib. methoda
number avg.
weight avg.
peak
dispersity
A B C A B C A B C A B C A B C A B C A B C A B C A B C
1.81 1.70 2.10 2.56 2.44 2.68 2.39 2.30 2.67 3.15 3.13 3.27 4.28 4.51 4.12 5.21 5.86 4.79 5.98 7.14 5.42 5.26 6.00 5.31 6.75 8.60 6.27
2.16 2.10 2.26 2.58 2.46 2.68 3.02 3.12 3.08 4.05 4.64 3.89 4.90 5.86 4.69 5.69 7.06 5.28 6.20 7.78 5.60 6.71 9.93 6.56 7.75 11.26 7.14
1.77 1.64 2.06 2.56 2.44 2.68 1.80 1.68 2.09 2.48 2.36 2.64 3.55 3.54 3.45 4.92 5.36 4.42 6.14 7.40 5.49 6.52 8.14 6.23 7.22 9.69 6.46
1.19 1.23 1.08 1.01 1.01 1.oo 1.26 1.36 1.16 1.29 1.48 1.19 1.14 1.30 1.14 1.09 1.21 1.10 1.04 1.09 1.03 1.28 1.66 1.24 1.11 1.31 1.14
a A: NRC method; B: F537 method; C: HMC method. Choay Laboratories, France. Cuitpold-Pharma,Germany. Rhone-Poulenc, France. Kabi, Sweden. 'Abbott Laboratories, Chicago, IL. Upjohn, Kalamazoo, MI. Novo Nordisk, Denmark.
those obtained by the NRC calibration. The weight-average MWs determined with the F537 calibration for the pure substance aprosulate compare well with its known MW of 2.39 kDa. In the case of aprosulate, the F537-derived values agreed better with the known MW than did the NRC-derived values. Substances with weight-average MW values of '8 kDa are shown in Table 2. Statistically significant differences between the methods were observed for the weight-average and peak MWs. Weight-average MWs calculated for this group with the F537 calibrant were higher than those obtained by either the NRC or HMC methods (p = 0.06 and 0.010, respectively). Similarly, the peak M W was also higher when calculated with the F537 calibrant (p = 0.014 and 0.016, respectively). Whereas both the number-average MW and the dispersity values were higher with the F537 calibrant, neither was statistically significant. Calculation by independent ANOVA gave p = 0.062 for the dispersity values and 0.058 for the number-average MW. These results agree with the fact that the number-average MW values are the lowest of the MW parameters and are, therefore, better covered by F537, which is skewed towards the low MW end compared with HMC. Because dispersity values are derived from M W values, this explanation applies to them t 0 0 . l ~
Discussion Calibration of the HPSEC column with a series of narrowrange calibrators has worked effectively in our laboratories for determining the MW profiles of heparins and LMWHs. These calibrators are effective because their M w s are well characterized by LALLS techniques and, because they have been isolated from UFH, their chemical structures very closely resemble those of the substances being analyzed. The MWs 726 /Journal of Pharmaceutical Sciences Vol. 84, No. 6, June 1995
Table 2-Comparative Analyses of Heparins by Narrow Range Calibrator, F537, and Heparin Mass Calibrator Methods sample name and code Braun-Melsungenb Heparin SuleparoideC Sandozd Natural Heparin HMF, Heparin International Std. No. 4 Abbott Std.,P 12700 Heparin, 19000
Molecuiar Weight (kDa) calib. methoda number avg. weight avg. peak A B C A B C A B C A B C A B C A B C A B C
7.64 10.52 7.21 7.63 10.26 7.28 10.65 13.40 10.48 11.97 16.44 11.94 9.77 13.47 9.95 10.47 18.53 10.81 14.64 21.06 14.57
8.41 13.72 8.26 9.16 16.29 9.40 11.82 16.43 11.78 12.89 19.37 12.84 11.04 20.87 12.00 11.09 21.52 11.86 15.84 23.57 15.58
9.08 14.94 8.71 8.32 12.63 7.91 11.63 16.02 11.74 12.30 17.78 12.42 11.05 21.41 11.86 11.05 18.88 10.68 14.23 21.03 14.33
dispersity 1.10 1.30 1.15 1.20 1.59 1.29 1.11 1.23 1.12 1.08 1.18 1.08 1.19 1.55 1.21 1.19 1.16 1.10 1.08 1.12 1.07
a A: NRC method; B: F537 method; C: HMC method. Braun-Meisungen, Germany. Syntex, Argentina. Sandoz, Switzerland. Abbott Laboratories, Chicago, IL.
of the NRC calibrators have been further corroborated by SDSPAGE electroph~resis.~~ The preparation of these compounds is both complex and time consuming, therefore limiting the availability of these substances. Recently, van Dedem and Nielsen7 proposed a new calibration method for the determination of the MW profile of LMWHs based on a single calibrant that is easier to produce and, therefore, more readily available. This calibrator is made by partial enzymatic degradation of UFH. The stated range of this calibrant is 1.2-12 kDa,6 which would be adequate for LMWH. We previously reporteds that the range of the method with the calibrant F913B was more limited, with an effective upper limit extending only to -7.5 kDa. This conclusion seems equally valid for F537. The calibrant F537 gives weight-average MW that are high, even for figures of the order 6 kDa, which is well below the limit of 8 kDa for LMWHs.15 In another report,'l we showed that a new depolymerized heparin mixture produced by benzylation and alkaline hydrolysis of UFH was a more effective calibrator for determining the MW profile of LMWH preparations. In this study, we showed that the HMC calibrator also has an improved MW range, which may be applicable to the analysis of heparins. These observations have been confirmed by obtaining reproducible results with different batches of HMC. Comparison of the M W profiles obtained by UV detection of the two calibrators shows that there is a notable difference in their shape. The elution profile of F537 is skewed towards the low molecular end as in the case of F913B.8 A very large proportion of this calibrator is therefore of lower MW. The elution profile of HMC is more symmetrical and more closely resembles that of the LMWHs being analyzed. This observation would account for the variations in the MWs determined. This dissimilarity in the MW profile is presumably related to the extent of depolymerization of the two calibrators and not to the method of production. It could also be due to the subsequent fractionation of the calibrators. As a consequence of this observation, the calibration range of HMC is extended to include higher MW heparins.
Therefore, F537 is not as optimal a calibrant for the determination of the MW profiles of LMWHs as HMC. F537 gives better results in the very LMW region, whereas above 3 kDa, HMC is superior and gives better values into the UFH reigon. In conclusion,based on our results, additional calibration mixtures can be developed for the determination of MW profiles of heparins and/or LMWHs.
References and Notes 1. Mardiguian, J. In Low Molecular Weight Heparins in Clinical Practice; Doutremepuich, C., Ed.; Marcel Dekker: New York, 1992; pp 7-12. 2. Casu, B. In Heparin. Chemical and Biological Pro erties, Clinical Applications; Lane, D. A.; Lindahl, U.; Eds.; CR& Boca Raton, FL, 1989; pp 25-49. 3. Barrowcliffe, T. W.; Mulloy, B.; Johnson, E. A.; Thomas, D. P. J.Pharm. Biomed. Anal. 1989, 7 , 217-226. 4. Nieduszynski, I. In Heparin. Chemical and Biological Properties, Clinical Applications; Lane, D. A.; Lindahl, U., Eds.; CRC: Boca Raton, FL, 1989; pp 51-63.
5. Emaneule R. M. Ph.D. Dissertation; Loyola University Medical Center, Maywood, IL, 1987. 6. Kuo, C.; Provder, T. In Detection and Data Analysis in Size Exclusion Chromatography;Provder, T., Ed.; ACS: Washington, DC, 1987; pp 2-28. 7. van Dedem, G.; Nielsen, J . I. Pharmeuropa 1991,3, 202-218. 8. Ahsan, A.; Jeske, W.; Fareed, J. Semin. Thromb. Hemostasis 1993,19 (Suppl 11, 63-68. 9. Jeske, W.; Ahsan, A.; Fareed, J. J . Thromb. Res. 1993, 70, 3950. 10. Ahsan, A.; Jeske, W.; Mardiguian, J.; Fareed, J . J . Pharm. Sci. 1994,83, 197-201. 11. Nielsen, J. I. Thromb. Haemostasis 1992, 68, 478-480. 12. Johnson, J. E.; Mulloy, B. Carbohydr. Res. 1976,51, 119-127. 13. Balke, S. T. In Detection and Data Analysis in Size Exclusion Chromatography;Provder, T., Ed.; ACS: Washington, DC, 1987; pp 202-219. 14. Edens, R. E.; Al-Hakim, A. L.; Weiler, J . M.; Rethwisch, D. G.; Fareed, J.; Linhardt, R. J. J . Pharm. Sci. 1992, 81, 823-827. 15. Low Molecular Mass Heparins: Draft monograph. Pharrneuropa 1991,3, 161-165. JS940602E
Journal of Pharmaceutical Sciences / 727 Vol. 84, No. 6, June 1995
Heparins are polymeric agents consisting of sulfated polysaccharide chains with broad molecular size dispersi0n.l The molecular weight (MW) of these chains ranges from 3 to 25 Low molecular weight heparins (LMWHs), formed either by chemical or enzymatic depolymerization of heparins, have MWs ranging from 3 to 8 kDa.l Molecular size is important in the determination of biological activity of any particular h e ~ a r i n ,so ~ various .~ methods have been developed to estimate this property. These techniques include measurement of osmotic pressure, sedimentation-diffusion, viscosity, and light-scattering behaviors.* Among these methods are the gel filtration (used here) and the viscometric procedures, which, however, must be calibrated by ultracentrifugal and light scattering method^.^ The determination of the MW of heparin preparations by high-performance size-exclusion chromatography (HPSEC) requires that the column(s) employed be properly calibrated with like molecules. In our laboratory, 19 oligosaccharide fractions isolated from native heparin are used to generate a log MW versus retention time calibration curve. These fractions are narrow MW distribution standards with dispersities of The oligosaccharide fractions cover a MW range of 1.3to 22.5 making them applicable for the MW profile determination of LMWHs and unfractionated heparins (UFHs). The MWs of these calibrators were determined by low-angle laser light scatter (LALLS) among other methods. Generation of such well-characterized polymer standards requires extensive laboratory time and is the main disadvantage of HPSEC methods6 A new procedure for obtaining the calibration curve for the HPSEC determination of the MW profile of LMWHs has been proposed by van Dedem and N i e l ~ e n .In ~ this procedure, a single polycomponent mixture (F913B) of heparin oligosaccharide chains, produced by heparinase-catalyzed partial < l . l . 5 3 6
@
Abstract published in Advance ACS Abstracts, April 1, 1995.
724 /Journal of Pharmaceutical Sciences Vol. 84, No. 6, June 1995
degradation of UFH, is used to determine calibration values. F913B is a product of partial heparinase digestion of UFHs. Because of the specificity of heparinase, the elution profile of F913B contains a series of regularly spaced peaks in the lower MW regions. Each of these peaks differs from its adjacent peak by one disaccharide unit, which has an assumed MW of 600 kDa. We have recently shown that this elution profile gives estimates of the MWs of various LMWHs that are in agreement with those determined by the NRC method, provided that the components of the LMWH mixture are <7.5 kDa. This procedure may be acceptable for measuring LMW components in the LMWHs [however, another calibration method would be required to measure the medium, (MMW) and high molecular weight (HMW) entities] as well as for obtaining MW profiles of UFHs. The percentages of HMW components determined by this method were consistently overestimated. Several other calibrators have since been developed for use with the technique of van Dedem and N i e l ~ e n .Ahsan ~ et d.l0 recently reported another calibrator produced by partial chemical degradation of the benzyl ester of UFH. The elution profile of this calibrator is considerably different from that of F913B obtained by the heparinase degradation method.1° The new calibrator, tentatively named HMC, gives more accurate MW parameters and dispersities. A second heparinasedegraded calibrator, F537, has also been proposed for determining the MW profile of heparins.ll A comparison of the elution profiles of F913B and F537 shows that the latter exhibits somewhat larger absorbances a t lower retention times and, therefore, contains more of the HMW material. Presumably, F537 was made by letting heparinase degradation run for a shorter time than in the case of F913B. In this paper, we compare the MW and dispersity data obtained for LMWHs and UFHs with the two newer calibrators, HMC and F537, with the data obtained by the reference narrow-rangecalibration (NRC) method used in our laboratories.
Experimental Section Calibration curves were prepared by running the appropriate compounds on a Waters 845 GPC-HPLC system (Millipore-Waters, Lexington, MA) equipped with Expert Ease software designed for polymer analysis. Each calibrator was dissolved in 0.5 M NaZS04 to give a lO-mg/mL solution. The HPLC system consisted of a VAX 3100 computer (Digital Corp.), a L A C E interface module, two 510 HPLC pumps, a 712 WISP autoinjector, a R401 differential refractometer, and a 484 tunable absorbance detector (all from Waters). The UV and the refractive index (RI) detectors were linked in series, with the outlet end of the columns attached to the UV detector. A 2O-pL aliquot of each sample was injected onto a joint column, TSK G3000SW and TSK G2000SW (Tosoh Corp., Japan), connected in that order. The protocol used for the runs with the F537 sample was the same as previously described,l' except that the aforementioned TSK columns were used instead of the Waters PAC I 60 and I 125 columns used in the other study. The method of acquiring the calibration curve data was very similar t o the one described in a previous paper.7 Both procedures used the TSK columns just mentioned, the differences being in the concentration of the mobile phase (0.5 M in our study and 0.2 M in the vanDedem and Nielsen
0022-3549/95/3184-0724$09.aO/O
0 1995, American Chemical Society and American Pharmaceutical Association
study7)and in the injection volume (20 pL in our study and 25 pL in the vanDedem and Nielsen study7). The flow rate for the mobile phase (0.5 M NazS04) was 0.5 m u m i n and the run time was 65 min. The UV determinations were made at 234 nm, and all analyses were done a t room temperature. This protocol was also followed for runs with HMC. For the partially degraded calibrators HMC and F537, MW assignments for each peak were made according to the method of van Dedem and N i e l ~ e n .Briefly, ~ the strengths of the RI and UV signals of each peak in the calibration mixture were determined. The ratio of these two values was then related to MW based on the assumption that the second-to-the-last peak of each mixture is that of a tetrasaccharide with a MW of 1.2 kDa, or the third one is a hexasaccharide (1.8 m a ) , and so on until each peak was assigned a MW. The criterion for selection of a particular peak as a reference point is that it agrees best with the rest of the peak ratios. The F537-derived calibration curves were prepared assuming the fourth peak from the low molecular end to be that of the octasaccharide. This assumption was in accord with that of Nielsen.l1 For the 19-standard-NRC method, the UV detector was set at 205 nm, where it is most sensitive, and the log MW of each calibrant was plotted against the retention time. A third-order regression equation was fitted to the data points with the Expert Ease software. Analyses of heparins and LMWHs used in this study were performed in the same way as for the calibrators. Following each run, the elution profile of each sample was analyzed by the three calibration methods. In addition to calculating the weight-average MW of each sample, the software was used to determine several other MW parameters based on the elution profile, including numberaverage MW, peak MW, and dispersity. The heparins and LMWHs used in this study are all sodium salts, supplied as white powders. The majority of these substances are marketed as injectables by pharmaceutical firms listed in the tables. For statistical analysis, MW data was separated into two groups ('8.0 kDa and '8.0 kDa). Data obtained for the various parameters by the three calibration methods were compared by independent analysis of variance (ANOVA). For each MW group,^ values of x0.05 were considered statistically significant (Primer of Biostutistics. Computer Program; McGraw-Hill: San Diego, CA). However, as HPSEC is not a n absolute method of MW determinations, the uncertainty expected in typical determinations of MW parameters will also vary with the method used to obtain the MW of the reference calibration standards. For instance, with LALLS, a popular absolute method of MW determination, a 10%uncertainty may be e ~ p e c t e d . ~
Figure 1-The comparative UV elution profiles of the (A) HMC and (B) F537 calibrator. The x-axis represents the elution time in min, the y-axis denotes the detector response in percentage (%), and the maximum signal in each case is given the value 100.
5.0
I
NRC v HMC V F537
I
20' 25
30
35
40
45
50
Retention time (min)
Results
Figure 2-Calibration curves generated by (0)the NRC method, (v)the F537 calibration method, and (v) the HMC method. Each of the calibration curves has a correlation coefficient of 9.999. The x-axis denotes the elution time in min, and the y-axis denotes the log MW.
Like F913B, F537 is prepared by depolymerization with heparina~e.~a~Jl The enzymatic action partially cleaves 1,4bonds between sulfated glucosamine and iduronic acid residues, the latter producing chains containing 4,Fi-unsaturated uronic acid groups at the nonreducing end, which absorb strongly a t 234 nm in the UV region. Depending on the conditions, a greater or smaller fraction of the bonds will be split. The fragments produced are multiples of disaccharide units, a consequence of the alternating uronic acid-glucosamine structure of heparin. A comparison of the HPSEC elution profiles of F537 and F913B shows that the former has somewhat more of the longer polymer chains than F913B. HMC is produced by benzylation and alkaline hydrolysis of UFH1s8 obtained from porcine intestinal mucosa. This method of degradation also leads to the formation of 4,5unsaturated uronic acid end groups, which absorb strongly a t 234 nm. The elution profile of this calibrator at 234 nm is considerably different from that obtained by heparinase degradation (Figure 1).Much larger absorbances are seen for HMC prior to an elution time of 35 min relative to the absorbances in the same region with the elution profile of F537. This result indicates that there is a larger amount of material a t the HMW ends of the profiles for HMC than for F537. The NRC calibrators are described in a previous publication.10 Briefly, they were made by fractionation of heparin
on Ultrogel AcA 44 agarose acrylamide (LKB-ProducteurAB) gel-permeation chromatographic c01umns.~This method is based on the procedure of Johnson and Mulloy.lZ Each of the individual narrow peaks in Figure 1 gives its retention time on the x-axis and the logarithm of the MW on the y-axis, giving the calibration curve. The best-fit calibration curves produced with each calibrator, the retention times of the individual narrow range standard runs giving the NRC curve, and the regularly spaced peaks of the F537 and HMC runs giving their respective curves are shown in Figure 2. All the curves had a correlation coefficient of >0.999. The curves are not identical, however. The HMC and NRC calibration curves are nearly identical at retention times between 36 and 45 min. This retention time corresponds to a MW between 2.5 and 8 kDa, the region where most LMWH components elute. The calibration curve of F537 is somewhat different. Computer software application of the three calibration curves so prepared to the elution profiles of the samples listed in the tables gave the molecular parameters. Table 1contains the MW profile of several compounds with a weight-average MW of 4 . 0 kDa. Statistically significant differences were observed only in the dispersity values, with those calculated by the F537 method being higher than those obtained by the NRC method 0, = 0.045) or the HMC method (p = 0.011). In the very LMW region (i.e., the first three entries of Table l), there is agreement between the F537-derived values and Journal of Pharmaceutical Sciences / 725 Vd. 84, No. 6,June 1995
Table 1-Comparative Analyses of LMWHs by Narrow Range Calibrator, F537, and Heparin Mass Calibrator Methods sample name and code CY 231 PQ2ZH AprosulateC CY 222b PIlWH Enoxaparind Fraxiparinb Flagmine Abbott' Std., 6200 Heparing Lot E, Upjohn Novoh 6076-4
Molecular Weight (kDa)
calib. methoda
number avg.
weight avg.
peak
dispersity
A B C A B C A B C A B C A B C A B C A B C A B C A B C
1.81 1.70 2.10 2.56 2.44 2.68 2.39 2.30 2.67 3.15 3.13 3.27 4.28 4.51 4.12 5.21 5.86 4.79 5.98 7.14 5.42 5.26 6.00 5.31 6.75 8.60 6.27
2.16 2.10 2.26 2.58 2.46 2.68 3.02 3.12 3.08 4.05 4.64 3.89 4.90 5.86 4.69 5.69 7.06 5.28 6.20 7.78 5.60 6.71 9.93 6.56 7.75 11.26 7.14
1.77 1.64 2.06 2.56 2.44 2.68 1.80 1.68 2.09 2.48 2.36 2.64 3.55 3.54 3.45 4.92 5.36 4.42 6.14 7.40 5.49 6.52 8.14 6.23 7.22 9.69 6.46
1.19 1.23 1.08 1.01 1.01 1.oo 1.26 1.36 1.16 1.29 1.48 1.19 1.14 1.30 1.14 1.09 1.21 1.10 1.04 1.09 1.03 1.28 1.66 1.24 1.11 1.31 1.14
a A: NRC method; B: F537 method; C: HMC method. Choay Laboratories, France. Cuitpold-Pharma,Germany. Rhone-Poulenc, France. Kabi, Sweden. 'Abbott Laboratories, Chicago, IL. Upjohn, Kalamazoo, MI. Novo Nordisk, Denmark.
those obtained by the NRC calibration. The weight-average MWs determined with the F537 calibration for the pure substance aprosulate compare well with its known MW of 2.39 kDa. In the case of aprosulate, the F537-derived values agreed better with the known MW than did the NRC-derived values. Substances with weight-average MW values of '8 kDa are shown in Table 2. Statistically significant differences between the methods were observed for the weight-average and peak MWs. Weight-average MWs calculated for this group with the F537 calibrant were higher than those obtained by either the NRC or HMC methods (p = 0.06 and 0.010, respectively). Similarly, the peak M W was also higher when calculated with the F537 calibrant (p = 0.014 and 0.016, respectively). Whereas both the number-average MW and the dispersity values were higher with the F537 calibrant, neither was statistically significant. Calculation by independent ANOVA gave p = 0.062 for the dispersity values and 0.058 for the number-average MW. These results agree with the fact that the number-average MW values are the lowest of the MW parameters and are, therefore, better covered by F537, which is skewed towards the low MW end compared with HMC. Because dispersity values are derived from M W values, this explanation applies to them t 0 0 . l ~
Discussion Calibration of the HPSEC column with a series of narrowrange calibrators has worked effectively in our laboratories for determining the MW profiles of heparins and LMWHs. These calibrators are effective because their M w s are well characterized by LALLS techniques and, because they have been isolated from UFH, their chemical structures very closely resemble those of the substances being analyzed. The MWs 726 /Journal of Pharmaceutical Sciences Vol. 84, No. 6, June 1995
Table 2-Comparative Analyses of Heparins by Narrow Range Calibrator, F537, and Heparin Mass Calibrator Methods sample name and code Braun-Melsungenb Heparin SuleparoideC Sandozd Natural Heparin HMF, Heparin International Std. No. 4 Abbott Std.,P 12700 Heparin, 19000
Molecuiar Weight (kDa) calib. methoda number avg. weight avg. peak A B C A B C A B C A B C A B C A B C A B C
7.64 10.52 7.21 7.63 10.26 7.28 10.65 13.40 10.48 11.97 16.44 11.94 9.77 13.47 9.95 10.47 18.53 10.81 14.64 21.06 14.57
8.41 13.72 8.26 9.16 16.29 9.40 11.82 16.43 11.78 12.89 19.37 12.84 11.04 20.87 12.00 11.09 21.52 11.86 15.84 23.57 15.58
9.08 14.94 8.71 8.32 12.63 7.91 11.63 16.02 11.74 12.30 17.78 12.42 11.05 21.41 11.86 11.05 18.88 10.68 14.23 21.03 14.33
dispersity 1.10 1.30 1.15 1.20 1.59 1.29 1.11 1.23 1.12 1.08 1.18 1.08 1.19 1.55 1.21 1.19 1.16 1.10 1.08 1.12 1.07
a A: NRC method; B: F537 method; C: HMC method. Braun-Meisungen, Germany. Syntex, Argentina. Sandoz, Switzerland. Abbott Laboratories, Chicago, IL.
of the NRC calibrators have been further corroborated by SDSPAGE electroph~resis.~~ The preparation of these compounds is both complex and time consuming, therefore limiting the availability of these substances. Recently, van Dedem and Nielsen7 proposed a new calibration method for the determination of the MW profile of LMWHs based on a single calibrant that is easier to produce and, therefore, more readily available. This calibrator is made by partial enzymatic degradation of UFH. The stated range of this calibrant is 1.2-12 kDa,6 which would be adequate for LMWH. We previously reporteds that the range of the method with the calibrant F913B was more limited, with an effective upper limit extending only to -7.5 kDa. This conclusion seems equally valid for F537. The calibrant F537 gives weight-average MW that are high, even for figures of the order 6 kDa, which is well below the limit of 8 kDa for LMWHs.15 In another report,'l we showed that a new depolymerized heparin mixture produced by benzylation and alkaline hydrolysis of UFH was a more effective calibrator for determining the MW profile of LMWH preparations. In this study, we showed that the HMC calibrator also has an improved MW range, which may be applicable to the analysis of heparins. These observations have been confirmed by obtaining reproducible results with different batches of HMC. Comparison of the M W profiles obtained by UV detection of the two calibrators shows that there is a notable difference in their shape. The elution profile of F537 is skewed towards the low molecular end as in the case of F913B.8 A very large proportion of this calibrator is therefore of lower MW. The elution profile of HMC is more symmetrical and more closely resembles that of the LMWHs being analyzed. This observation would account for the variations in the MWs determined. This dissimilarity in the MW profile is presumably related to the extent of depolymerization of the two calibrators and not to the method of production. It could also be due to the subsequent fractionation of the calibrators. As a consequence of this observation, the calibration range of HMC is extended to include higher MW heparins.
Therefore, F537 is not as optimal a calibrant for the determination of the MW profiles of LMWHs as HMC. F537 gives better results in the very LMW region, whereas above 3 kDa, HMC is superior and gives better values into the UFH reigon. In conclusion,based on our results, additional calibration mixtures can be developed for the determination of MW profiles of heparins and/or LMWHs.
References and Notes 1. Mardiguian, J. In Low Molecular Weight Heparins in Clinical Practice; Doutremepuich, C., Ed.; Marcel Dekker: New York, 1992; pp 7-12. 2. Casu, B. In Heparin. Chemical and Biological Pro erties, Clinical Applications; Lane, D. A.; Lindahl, U.; Eds.; CR& Boca Raton, FL, 1989; pp 25-49. 3. Barrowcliffe, T. W.; Mulloy, B.; Johnson, E. A.; Thomas, D. P. J.Pharm. Biomed. Anal. 1989, 7 , 217-226. 4. Nieduszynski, I. In Heparin. Chemical and Biological Properties, Clinical Applications; Lane, D. A.; Lindahl, U., Eds.; CRC: Boca Raton, FL, 1989; pp 51-63.
5. Emaneule R. M. Ph.D. Dissertation; Loyola University Medical Center, Maywood, IL, 1987. 6. Kuo, C.; Provder, T. In Detection and Data Analysis in Size Exclusion Chromatography;Provder, T., Ed.; ACS: Washington, DC, 1987; pp 2-28. 7. van Dedem, G.; Nielsen, J . I. Pharmeuropa 1991,3, 202-218. 8. Ahsan, A.; Jeske, W.; Fareed, J. Semin. Thromb. Hemostasis 1993,19 (Suppl 11, 63-68. 9. Jeske, W.; Ahsan, A.; Fareed, J. J . Thromb. Res. 1993, 70, 3950. 10. Ahsan, A.; Jeske, W.; Mardiguian, J.; Fareed, J . J . Pharm. Sci. 1994,83, 197-201. 11. Nielsen, J. I. Thromb. Haemostasis 1992, 68, 478-480. 12. Johnson, J. E.; Mulloy, B. Carbohydr. Res. 1976,51, 119-127. 13. Balke, S. T. In Detection and Data Analysis in Size Exclusion Chromatography;Provder, T., Ed.; ACS: Washington, DC, 1987; pp 202-219. 14. Edens, R. E.; Al-Hakim, A. L.; Weiler, J . M.; Rethwisch, D. G.; Fareed, J.; Linhardt, R. J. J . Pharm. Sci. 1992, 81, 823-827. 15. Low Molecular Mass Heparins: Draft monograph. Pharrneuropa 1991,3, 161-165. JS940602E
Journal of Pharmaceutical Sciences / 727 Vol. 84, No. 6, June 1995