CA2710585A1 - An adsorptive purification method for iodixanol - Google Patents

Abstract

This invention relates generally to non-ionic X-ray contrast agents. It further relates to a method of using solid adsorbents to reduce backpeaks content in the purification of iodixanol. In particular, it relates to the use of non-polar organic adsorbents with the average pore diameter smaller than about 30 nm. Specific examples of adsorbents of the instant invention include non-polar acrylic ester, di-vinyl benzene resins, poly-styrene di-vinyl benzene resins, and carbon adsorbents. In certain embodiments, upwards of 30%
of the backpeak levels and 60% N-acetyl cyclic iodixanol levels are reduced for a 5%
loss of iodixanol.

Description

An Adsorptive Purification Method For Iodixanol CROSS REFERENCE TO RELATED APPLICATIONS

The present application claims benefit of priority under 35 U.S.C. 119(e) to United States Provisional Application number 61/227,103 filed July 21, 2009, the entire disclosure of which is hereby incorporated by reference.

TECHNICAL FIELD

This invention relates generally to non-ionic X-ray contrast agents. It further relates to a method of using solid adsorbents to reduce backpeaks content in the purification of iodixanol. In particular, it relates to the use of non-polar organic adsorbents with the average pore diameter smaller than about 30 nm.

BACKGROUND OF THE INVENTION

Iodixanol is a non-ionic X-ray contrasting agent produced in large quantities by GE
Healthcare in Lindesnes, Norway. The industrial production of iodixanol involves a multistep chemical synthesis. The lowest yield is in the final step, involving coupling of the intermediate 5-acetamido-N,N'-bis(2,3-dihydroxypropyl)-2,4,6-triiodoisophthalamide ("Compound A") with epichlorohydrin under basic conditions. Yield losses are attributed to side reactions leading to byproducts ("backpeaks"). These backpeaks must be minimized to provide iodixanol that meets regulatory requirements like those mandated by US

Pharmacopeia.

While the purification of iodixanol is necessary and extensive, it is also important that such process does not lead to substantial loss of iodixanol. There exists a need for economic procedures to achieve the desired purity of iodixanol, while minimizing the loss of the final product during the purification process.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. I illustrates iodixanol purification in aqueous solution with various mostly non-polar adsorbents.

FIG. 2 illustrates aqueous Iodixanol solution backpeak reduction from control for best non-polar adsorbents.

FIG. 3 illustrates aqueous iodixanol solution N-acetyl cyclic iodixanol reduction using mostly non-polar adsorbents.

FIG. 4 illustrates backpeak adsorption isotherms from aqueous solution for resin and BGHHM and NCOI-125 carbons.

FIG. 5 illustrates iodixanol adsorption isotherms from aqueous solution for resin and BGHHM and NCO 1-125 carbons FIG. 6 illustrates average performance of adsorbents for purifying iodixanol in a 20.8 wt% aqueous solution, showing the percentage of the original mass adsorbed by the adsorbent.

FIG. 7 illustrates average separation factors for iodixanol to backpeaks and iodixanol to N-acetyl cyclic iodixanol in aqueous solutions at 25 C

FIG. 8 illustrates cumulative pore volume as a function of average pore size for BGHHM Carbon.

SUMMARY OF THE INVENTION

The present invention provides a large scale purification process of iodixanol. This process involves the use of non-polar organic adsorbents in the treatment of bulk iodixanol, where the average pore diameter is smaller than about 30 nm. Specific examples of adsorbents of the instant invention include non-polar acrylic ester, di-vinyl benzene resins, poly-styrene di-vinyl benzene resins, and carbon adsorbents. In some preferred embodiments, adbsorbents with 90% of the pore volume smaller than 30 nm are employed.
In other embodiments, adsorbents with surface area for adsorption about 1350 m2/g of adsorbent are used.
The less polar impurities include N-acetyl cyclized iodixanol, cyclized iodixanol, and primary o-alkyl trimer. In certain embodiments, upwards of 30% of the backpeak levels and 60% N-acetyl cyclic iodixanol levels are reduced for a 5% loss of iodixanol.
For example, upwards of 30% of the backpeak levels reductions may be obtained for a 2:1 solids to adsorbent mass ratio. In another example, separation factors of four may be obtained between iodixanol and backpeaks in adsorptive purification, with about a 20%
reduction in backpeaks for a 5% loss of iodixanol to the adsorbed phase.

OH

O IH
OH N OH
N
I / I OH OH
+ OH-O \ N ~ O \ N

OH OH
OH OH
Intermediate lohexol HO N N~ ' N N OH
OH H I / / H OH Intermediate OH-HO'Y` H 0 0 N I OH
OH OH
lodixanol HO^y N \ NI v vN N OH
I H I H
O H OH

HOs` H 0 0 N /Y\ O
OH OH OH
O
OH
Primary o-alkyl timer N
/ I I OH
p NH

OH
OH

o OH 'Yo 1 o O I olr O I --rN N O
\ ~ \ HO OH H I O
~\ ~\ OH N O H N N
H OH OH OH- 1 u H ~~OH
OH 1 I I I ~~ HO "Y'N O 1 / I OH
HO "-rN 0 0 H N ~\OH OH 0 N /--T-'oH
OH lodixanol OH N-acetyl cyclized H on iodixanol H
N O

OH N N ~OH
OH
H I / I
HO O
OH
O N\OH
H
Cyclized iodixanol OH

Scheme 1 DETAILED DESCRIPTION OF THE INVENTION

The instant invention is directed to purifying bulk iodixanol using non-polar organic adsorbents, where the average pore diameter is smaller than about 30 nm. In a preferred embodiment, analyses on iodixanol solutions before and after treatment with the adsorbents are performed by High Performance Liquid Chromatography (HPLC). For example, based on a HPLC assay, backpeaks as the major contaminants may be at 0.80-0.90 wt%.
In some embodiments, water is chosen as the solvent for iodixanol in order to exploit the more non-polar nature of the backpeaks relative to iodixanol. This is to create a very polar interaction of iodixanol with water, and a more hydrophobic environment on the adsorbent, to selectively remove backpeaks. This strategy is consistent with the results of reversed phase HPLC conditions, where backpeaks are selectively adsorbed to a greater extent (i.e. have longer retention times than iodixanol) onto a non-polar C 18 column when eluting a primarily aqueous mobile phase.
Among the preferred adsorbents selected were the non-polar acrylic ester and di-vinyl benzene resins, as well as a set of poly-styrene di-vinyl benzene resins, and a number of carbon adsorbents. For contrast, a set of polar silica adsorbents was also studied. A 20.8 wt% solids aqueous iodixanol solution was contacted with the adsorbents at 25 C in a mass ratio of 2:1 iodixanol to adsorbent in order to see if there was a dramatic effect on iodixanol purity with the high loading of adsorbent (Figure 1).
The data show that certain types of carbon and acrylic ester resin adsorbents purify the iodixanol from the less polar backpeaks and N-acetyl cyclic iodixanol in aqueous solution. The amount of backpeak reduction is shown in Figure 2 for the three carbons and the acrylic ester resin that performed the best in purifying the iodixanol.
These four adsorbents reduced the backpeak level by 30% or more and are all commercially available.
In addition, cyclic iodixanol is also reduced by 65-70% as shown in Figure 3.

Although the CG71 M has a lower surface area (500m2/g) than BGHHM Carbon (1350 m2/g), one important comparison for adsorbents is capacity for the desired material to adsorb relative to the product. The preferred way to test this is to perform isotherm studies.
To measure adsorption isotherms, known weights of dry adsorbent were added to aliquots of the standard aqueous iodixanol solution of 20.8 wt% solids in mass ratios of 4:1 up to 30:1 iodixanol to adsorbent. The slurries were contacted with mixing for 24 hours, and then aliquots of the resulting supernatant liquid were analyzed for iodixanol, backpeak and N-acetyl cyclic iodixanol content by the standard HPLC analysis.
The results for backpeak adsorption as a function of *concentration are shown in Figure 4. The data are plotted as amount of backpeak adsorbed per mass of adsorbent versus the mass concentration of backpeak in the solution at equilibrium.
The scatter in Figure 4 is due to several factors. The measurements are done by dilution of 200 micro-liters of the supernatant liquid with water and then injection into the HPLC. Although the masses are recorded to four decimal places, there are still potential inaccuracies associated with taking a small sample of the solution. In addition, to get a measured mass concentration from the HPLC, the area for the peak from the HPLC
UV
measurement at 254 nm is correlated for the control samples, knowing the concentration of each component in the control. The area response of the adsorbed samples is then taken as a ratio of it to the control in order to determine the mass concentration of the sample. In addition, there could be some imprecision involved in the manual integration of the peaks.
Despite the scatter, the data show that BGHHM has a better capacity for backpeaks than NC01-125.

Similar results are shown for iodixanol adsorption as a function of loading.
Figure 5 shows that all adsorbents adsorb only fairly small amounts of iodixanol relative to the solution concentration of nearly 20-wt%.
To examine the isotherm data, the average performance of each adsorbent may be plotted. Because the overall shape of the isotherms over these concentration ranges is fairly flat, the average will provide some relative performance of the adsorbent at purifying iodixanol. Such a plot is shown in Figure 6. The performance is based on the percentage of the original mass in the solution that is adsorbed by the adsorbent. The standard deviations are also given, and are large due to the nature of the average measurement being done over a fairly wide concentration range. On average, all three adsorbents remove about 15-25% of the initial backpeak level (low concentration), while adsorbing only 4-7% of the iodixanol (high concentration). This indicates that the adsorbents are highly selective for the backpeaks relative to the iodixanol.

From the measurements, an average separation factor can also be calculated.
The separation factor is defined as the ratio of the mole fractions of components i and j in phases I and 2. See King, C.J. Separation Processes. 1980: McGraw-Hill.

X11 l xji ail x;2 l x;2 Eq. 1 For aij>1, component i will be concentrated in phase 1, and component j in phase 2. In this case, iodixanol is component i and backpeak or N-cyclized is component j, with phase 1 being the solution and phase 2 being the solid adsorbent. As can be seen in Figure 7, the average separation factors for iodixanol to backpeaks and iodixanol to N-acetyl cyclic iodixanol are shown. All are greater than one, indicating that iodixanol is concentrated in the solution phase, and the backpeaks and N-acetyl cyclic iodixanol are concentrated in the adsorbed phase. The larger the positive number, the better the separation. The N-acetyl cyclic iodixanol is less polar and thus separated better from the iodixanol than the backpeaks.
All three non-polar adsorbents are about equal at backpeak separation, with the CG71 M and the NCO 1-125 being better at N-acetyl cyclic iodixanol separation. This separation factor comparison may be a preferred method of comparing adsorption performance over the range of concentrations than the isotherms, due to the potential inaccuracies in calculations noted above.
A study of Calgon BGHHM carbon pore size distribution (see Figure 8) was performed, and is shown to give an example of the pore sizes that contribute to the adsorptive purification of iodixanol. In this way, the preferred pore sizes for purification may be determined in order to optimize adsorbent performance.
From the cumulative pore volume for BGHHM carbon, it can be seen that 90% of the pore volume is less than 30 nm, making the adsorbent highly micro-porous. The surface area for adsorption is about 1350 m2/g. For comparison, the CG71M acrylic ester resin has an average pore diameter of 25 nm and a surface area of 500 m2/g, according to the manufacturer. Further characterization of all the promising adsorbents should be performed, to confirm that micro-porous adsorbents with a majority of the non-polar pores less than 30 nm are preferred adsorbents for separating the impurities from the iodixanol in polar solvents.
The invention is illustrated further by the following examples that are not to be construed as limiting the invention in scope to the specific procedures or products described in them.

EXAMPLES
EXAMPLE 1: Reaction conditions Reagents for adsorbent studies: 2-Methoxyethanol (2-ME; Aldrich, HPLC grade, >=99.9%, Lot#03758HE) was used without further purification. De-ionized water was provided by a Millipore Synergy 185 system, giving 18 mega-ohm water.
lodixanol solids were from Lindesnes Batch #10429846. Adsorbents used and their manufacturers are listed in Table 1.

Table 1 Adsorbent types and manufacturers used for iodixanol purification Adsorbent Manufacturer Activated Carbons GC 8x30S General Carbon Carbsorb 40 Calgon CPG Calgon OLC Calgon BL Calgon BG-HHM Calgon NCOI-125 Norit M-1311 Norit Polymers CG161M Acrylic Ester Rohm & Haas CG71 M Acrylic Ester Rohm & Haas XAD-7 Acrylic Ester Rohm & Haas XAD-4 Di-Vinyl Benzene Rohm & Haas XSD-2 Di-Vinyl Benzene Rohm & Haas Silicas C930 PQ Corporation D350EL PQ Corporation In addition, some other adsorbents were identified using the Supelco catalog, as potentially effective for these types of separations, and are listed in Table 2.

Table 2 Listing of various polar and non-polar adsorbents available from Supelco A Pare V Mean pore Adsortert Strpelco # nlbnufach e5- CAS # Con xisifion (5-r?/g) (nl/g) Sias (5-m) Particle Sb e St{ de e D &X-8 2027 S pdco Prrylic Ester 1 0.7 22 40-60 Mash C Tbigel (Prrb& e) Uldaivab 9d )(E-305 502537 Stow, RSd-I polystyrene 50-100M h P4EGB 42696- Sgxlco 9043-77 GB) 10. 300-1200 m Pb1y(4-hylsyrt co-P4EUB 42696- S.peloo 904377 IYB) 7.1,1 500Nrn PS-D B 4269& St.Qeloo 900370= Pbly(styrer co-DB) 4. 300-800 in Art,erliteXAD7hP 13361 R81-I 37380431PaylicEster 1.1 Q 2D.6)Mesh prTbetiite XAD16FP 1335r >L 90034694 PSGB 1.82 10. 20-60 MEsh Anterlite XAD1180 1 R&H 97396-5M 5CO
phend-fo dehyde/nrthyld Antertite XAD761 1035 R8iH FG G43 6Q 1650 Mesh PrTte'dTomOG761 1 RRFI Pcrylice5ter 1.1 25. 80-160 ~fn berdranOG161 1 R&H PSD õ 1,45 15 50-10O.pm An berdvern OG300 130 RS H PSGB 1. 30. 25-50 t m Arrtthr= OG1000S 13911, RSH PS-M 1. 100. 25-50 ni Diaion FR20 1 Mtabishi PS-GB 1. 26. 250-850 rrt Dixon FP-20SS 1361 Mtstbishi PSGB 1. 26 75-150 lm Dimon FP-2M3 13601 Mtstbishi PM4 1. 17. 25-50 Sepabeads SP-20SS 13617 Mts ishi PS.DvB 1.01 26 50500 Adsorption Experimental: Adsorption studies used a standard solution of iodixanol in 2-ME
or de-ionized water. Several controls of each solution were measured for iodixanol, backpeak and N-acetyl cyclics via HPLC. HPLC samples of solutions used in controls and adsorption experiments were prepared from a weighed 125- 1 aliquot diluted in a weighed amount of 25 ml de-ionized water.
Silica and carbon adsorbents were prepared by drying at 250 C and 200 mm Hg vacuum in an oven in order to remove any adsorbed species prior to mixing with the solution.
Polymer adsorbents were dried at 100 C and 200 mm Hg.
The adsorption was performed by weighing out 10 ml of standard solution into a ml jar. Next a weighed amount of adsorbent was added to the jar, along with a Teflon stir bar. The jar was then sealed under nitrogen and mixed on a stir plate for 24-hours. The temperature of the ambient air was also monitored, and fluctuated between 22-27 C.
After equilibration, a portion of the resulting solution was filtered using a 0.5 m Teflon syringe filter to remove any adsorbent particles. Then a 125-0 aliquot of this filtered solution was diluted in a weighed amount of 25 ml de-ionized water. This diluted sample was then analyzed via HPLC.

Subsequently, the areas for peaks of the known standard solutions were used to calculate the amount of iodixanol, backpeak and N-acetyl cyclics in the equilibrated adsorption solution. The dilution factor was taken into account, and the amounts of each component in the solution and adsorbed phases were calculated.
HPLC Analysis of Adsorbent Solutions: HPLC analysis was performed on an Agilent 1100 series HPLC as indicated below.
Column: YMC-Pack ODS-AM, S-5 m, 120A, 15 cm and diameter 4.6 mm, Mobile phase: - Millipore Synergy 185 system de-ionized water.
- 50 % Acetonitrile in Millipore Synergy 185 system de-ionized water.
Flow: 1.25 ml/min.

Detector: UV - 254 nm (15 l flow cell and 10 mm path length) Temperature: 25 C or ambient Gradient:

50% ACN in De-ionized Time (min) de-ionized water water 0.0 6 94 2.7 6 94 5.5 14 86 16.5 14 86 19.5 26 74 26.5 90 10 31.5 90 10 EXAMPLE 2: Silica adsorption using 2-methoxyethanol solution 157.5g of iodixanol (GE Healthcare Lot#1042986) was dissolved in 771.7 g of 2-Methoxyethanol (Aldrich Lot#03758HE) to give a 16.95 Wt% solids (iodixanol and impurities) in solution. 25 ml aliquots of this solution were added to a glass jar with a magnetic stir bar. All adsorbents were dried at 250 C and 250 mm Hg vacuum in a drying oven. Then a weighed amount of dry adsorbent was also added to the jars, except for the control, where no adsorbent was added. The amounts are summarized in Table 3, to give various iodixanol to adsorbent ratios. The glass jars were then placed on a magnetic stirring board and allowed to stir and equilibrate for 20 hours. The subsequent mother liquor was then filtered using a 2-micron Teflon syringe filter. A 125- l aliquot of this filtered solution was then diluted in 25 ml of 16 mega-ohm purified water and analyzed for the amount of iodixanol, backpeaks and N-acetyl cyclized solids remaining in the solution.
The relative amount of each component in each solution was then compared to the control to determine how effective the adsorbent was at removing the impurities. The data in Table 3 shows that the silica adsorbents are not effective at separating out the non-polar impurities from the iodixanol in 2-Methoxyethanol.

Table 3 Summary of Silica Adsorption Experiments using lodixanol dissolved in Methoxyethanol Notebock# AS32-78-1 AS326-78-2 AS326-78-3 AS32-78-4 AS323-788 PS32f-78-9 AS32f-Sdtbm Nffis () 25. 26.48 26.80 25.47 26.51 25.48 26.40 26.47 28.42 25.45 laizand Cadent () 4.4799 4.4884 45426 4.4867 4.4964 4.4884 44850 4.4867 4.4782 4.4833 Admbent Type None D350EL Slip D350EL Slip D350EL Silica 8100 Silica R10o Slica R100 Silica C930 Slica 0:930 Silica C930 Silica Adsaberd K/bss (g) - 0.44 0.91 224 225 0. 0.44 0.44 0.89 224 Ml3rx.fadua - PQ Cap, PQ Corp. PQ Corp. PQcap PQ Cap. PQ Corp PQ ccrp PQ Carp.
PQ OprP, Lct# - S009 9009 9009 R40507313 R40507313 R40507313 52642 52542, 5264 kxixand Mt~sstSdids Mffis - 10.20 4.99 200 2 5.04 10.1 10.20 5.03 200 Ong T (C) 250 250 250 250 250 250 250 250 Egcilitxiunhidos (Ntrraliaed %oorrpment in skim) locixand 99.1918 99.0714 99.1934 990780 99.2387 98. 99. 98.9975 99.0964 99.1139 Bafteaks 0.7548 0.8144 0.730 0.8084 0.70601 0.9206 0.8258 0. 0.8163 98106 zad 0.0535 0.1142 0.0717 0.1136 0.055 0.101 0.1050 0.1066 0.086900755 Example 3: Alumina adsorption using 2-methoxyethanol solution Using the same control solution prepared in Example 2, and the same method of drying the adsorbents, a similar set of experiments was done using Alumina adsorbent. The amounts are summarized in Table 4, to give various iodixanol to adsorbent ratios. The glass jars were then placed on a magnetic stirring board and allowed to stir and equilibrate for 20 hours. The subsequent mother liquor was then filtered using a 2-micron Teflon syringe filter.
A 125- l aliquot of this filtered solution was then diluted in 25 ml of 16 mega-ohm purified water and analyzed for the amount of iodixanol, backpeaks and N-acetyl cyclized solids remaining in the solution. The relative amount of each component in each solution was then compared to the control to determine how effective the adsorbent was at removing the impurities. The data in Table 4 show that the alumina adsorbents are not effective at separating out the non-polar impurities from the iodixanol in 2-Methoxyethanol, and in fact due to the acidic nature of the adsorbent surface, they catalyze the reaction of iodixanol to the N-acetyl cyclized impurity. The higher the amount of adsorbent added, the larger the N-acetyl cyclized impurity formed.

Table 4 Summary of Alumina Adsorption Experiments using iodixanol dissolved in Methoxyethanol Notebook # AS326-78-1 AS326-78-5 AS326-78-6 AS326-78-7 Solution Mass (g) 26.43 26.50 26.44 26.50 lodixanol Content 4.4799 4.4918 4.4816 4.4918 Adsorbent Type None Alumina 9139A Alumina 9139A Alumina 9139A
Adsorbent Mass (g) - 0.46 0.89 2.25 Manufacturer - UOP UOP UOP
Lot# - 4413000226 4413000226 4413000226 lodixanol Mass/Solids Mass - 9.76 5.04 2.00 Drying T C) - 250 250 250 Equilibrium Analysis (Normalized % component in solution) lodixanol 99.1918 98.1269 97.5749 96.6652 Back peaks 0.7548 0.7008 0.7425 0.6906 N-acetyl-c clized 0.0535 1.1723 1.6825 2.6443 Example 4: Polymer resin adsorption using 2-methoxyethanol solution Using the same control solution prepared in Example 2, and the same method of drying the adsorbents at 100 C and 250 mm Hg vacuum, a similar set of experiments was done using polymeric resin adsorbents. The amounts are summarized in Table 5, to give various iodixanol to adsorbent ratios. The glass jars were then placed on a magnetic stirring board and allowed to stir and equilibrate for 20 hours. The subsequent mother liquor was then filtered using a 2-micron Teflon syringe filter. A 125- 1 aliquot of this filtered solution was then diluted in 25 ml of 16 mega-ohm purified water and analyzed for the amount of iodixanol, backpeaks and N-acetyl cyclized solids remaining in the solution.
The relative amount of each component in each solution was then compared to the control to determine how effective the adsorbent was at removing the impurities. The data in Table 5 shows that the polymer adsorbents are not effective at separating out the non-polar impurities from the iodixanol in 2-Methoxyethanol. Again, due to the acidic nature of the adsorbent surface, some polymer resins catalyze the reaction of iodixanol to the N-acetyl cyclized impurity.

Table 5 Summary of Resin Adsorption Experiments Using lodixanol Dissolved in 2-Methoxyethanol Notebook # AS326-78-1 AS326-79-18 AS326-79-19 AS326-79-20 AS326-79-21 AS326-79-Solution Mass 26.43 26.54 26.47 26.43 26.36 26.40 26.45 26.50 lodixanol Content (g) 4.4799 4.4980 4.4862 4.4804 4.4675 4.4751 4.4824 4.4923 Adsorbent Type None Amberlite XAD-2 Amberlite XAD-2 A berlite XAD-4 Amberlite XAD-7 Dmex MSA IRC76 Ambeityst 15 Adsorbent Mass (g) - 2.005 0.408 2.003 2.002 2.003 2003 2.003 Manufacturer - Rohm & Haas Rohm & Haas Rohm & Haas Rohm & Haas Dow Rohm & Haas Rohm & Haas Lot# - 04917TZ 04917TZ 03803HF 03729DF 13107EA 6.22E+06 210126 lodixanol Mass/Solids Mass - 2.24 11.00 2.24 2.23 2.23 2.24 2.24 Drying T C 100 100 100 100 100 100 100 Equilibrium Analysis (Normalized % component in solution) Iodixanol 99.1918 98.2166 98.9811 98.5871 99.2723 99.1813 99.1963 99.0813 Back ks 0.7548 0.7480 0.7729 0.7597 0.6807 0.7596 0.7389 0.8711 N-ace ized 0.0535 1.0354 0.2460 0.6532 0.0470 0.0590 0.0648 0.0476 Example 5: Further Polymer Resin Adsorption using 2-Methoxyethanol Solution Using the same control solution prepared in Example 2, and the same method of drying the adsorbents at 40 C and 250 mm Hg vacuum, a similar set of experiments to Example 4 was done using new set of non-polar polymeric resin adsorbents. The amounts are summarized in Table 6, to give various iodixanol to adsorbent ratios. The glass jars were then placed on a magnetic stirring board and allowed to stir and equilibrate for 20 hours. The subsequent mother liquor was then filtered using a 2-micron Teflon syringe filter. A 125- l aliquot of this filtered solution was then diluted in 25 ml of 16 mega-ohm purified water and analyzed for the amount of iodixanol, backpeaks and N-acetyl cyclized solids remaining in the solution. The relative amount of each component in each solution was then compared to the control to determine how effective the adsorbent was at removing the impurities. The data in Table 6 shows that the polymer adsorbents are not effective at separating out the non-polar impurities from the iodixanol in 2-Methoxyethanol.

Table 6 Summary of Resin Adsorption Experiments using lodixanol dissolved in 2-Methoxyethanol Notebook # AS326-99-1 AS326-99-2 AS326-99-3 AS326-99-4 AS326-99-5 AS326-99-6 Solution Mass (g) 6.9212 6.7906 7.1947 7.1462 7.1619 7.1891 Iodixanol Content 1.1731 1.1510 1.2195 1.2113 1.2139 1.2186 Adsorbent Type None CG71M CG71M None CG161M CG161M
Adsorbent Mass (g) - 0.2302 0.6114 - 0.2428 0.6085 Manufacturer - Rohm & Haas Rohm & Haas - Rohm & Haas Rohm & Haas Lot # - 0003604478 0003604478 - 0003558506 0003558506 lodixanol Mass/Solids Mass - 5.000 1.995 - 5.000 2.003 Drying T (C) - 100 100 - 100 100 Equilibrium Analysis (Normalized % component in solution) lodixanol 99.1302 99.0317 99.0926 99.1604 99.2252 99.2095 Backpeaks 0.7859 0.8773 0.8397 0.7760 0.7153 0.7471 N-acetyl- dized 0.0838 0.0910 0.0677 0.0636 0.0595 0.0435 Example 6: Carbon Adsorption using Aqueous Solution 55.05g of iodixanol (GE Healthcare Lot# 1042986) was dissolved in 220.1 g of mega-ohm purified water to give a 20.76 Wt% solids (iodixanol and impurities) in solution.
ml aliquots of this solution were added to a glass jar with a magnetic stir bar. All 10 adsorbents were dried at 250 C and 250 mm Hg vacuum in a drying oven. Then a weighed amount of dry adsorbent was also added to the jars, except for the control, where no adsorbent was added. The amounts are summarized in Table 7, to give various iodixanol to adsorbent ratios. The glass jars were then placed on a magnetic stirring board and allowed to stir and equilibrate for 20 hours. The subsequent mother liquor was then filtered using a 2-micron Teflon syringe filter. A 125- l aliquot of this filtered solution was then diluted in 25 ml of 16 mega-ohm purified water and analyzed for the amount of iodixanol, backpeaks and N-acetyl cyclized solids remaining in the solution. The relative amount of each component in each solution was then compared to the control to determine how effective the adsorbent was at removing the impurities. The data in Table 7 shows that some carbon adsorbents with the right properties are effective at separating out the non-polar impurities from the iodixanol in a polar water solution.

Table 7 Summary of Carbon Adsorption Experiments using lodixanol dissolved in water Notebook# AS99-7 46998 AS939 ASW10 A599-11 AS99-12 AS99-13 A599.14 A S99-15 Souition Klass ( 10. 10.9528 10.9130 11.031 10.9681 11.0744 11.0208 10.9792 11.0233 lod xaid Content 22754 2.2759 2.2655 2.2902 2.2770 22990 2.2B79 2.2793 2.
A& "bent Type None Carbon &30S Cabscrb 40 CPG Carbon CLC Carom BL Carbon BGH-M
Cabin NC01-125 Carbon W1-1311 Cabin Pdmbent 11b% (g) - 1.138 1.1329 1.1452 1.1383 1.1518 1.1445 1.1402 1.1449 Mardaduer - General Caton CAgcn Calagon C@Igon Calgm Nail Norit Ld # - None- reseath None- resew th None- resecrrh None - mseard None- reseadi None - reseadi We - reseadl None - reseath lodixaio Mms/Sdids Wass - 1.9989 1.999 1.9999 2A004 1.9980 1. 1.9990 1.9988 Drying T (C) 250 250 250 250 250 250 250 Z-0 lxluilibriun N alysis (Nrxnel¾e 1 %ccnponert in solution) lociixa-d 99.221 99.2257 99.3391 99.3555 99.2463 99.4259 99.5267 99.517 99.5022 0.6662 0.7398 0.5853 0.6054 0. 0.5547 0.4382 0.4479 0.4704 N ia~d 0.1122 O.c3q 0.0756 0.0391 0.0702 0.0194 0.0351 0.034 0.027 Example 7: Polymer Resin and Polar Silica Adsorption using Aqueous Solution Using the same control solution prepared in Example 6, and the same method of drying the adsorbents at 100 C for polymer (40 C for CG series polymers) and 250 C for silica and 250 mm Hg vacuum, a similar set of experiments to Example 6 was done using new set of non-polar polymeric resin adsorbents and a set of polar silica for contrast. The amounts are summarized in Table 8, to give various iodixanol to adsorbent ratios. The glass jars were then placed on a magnetic stirring board and allowed to stir and equilibrate for 20 hours. The subsequent mother liquor was then filtered using a 2-micron Teflon syringe filter.
A 125- 1 aliquot of this filtered solution was then diluted in 25 ml of 16 mega-ohm purified water and analyzed for the amount of iodixanol, backpeaks and N-acetyl cyclized solids remaining in the solution. The relative amount of each component in each solution was then compared to the control to determine how effective the adsorbent was at removing the impurities. The data in Table 8 shows that some polymer adsorbents are effective at separating out the non-polar impurities from the iodixanol in water, while the polar silicas are not effective at iodixanol purification in a polar solvent like water.

Table 8 Summary of Resin & Silica Adsorption Experiments using lodixanol dissolved in water Notebook # AS99-23 AS99-16 AS99-17 AS99-18 AS99-19 AS99-20 AS99-21 AS99-22 Solution Mass 111108 10.9882 10.9293 11.0162 10.9395 10.9812 10.8965 10.9658 lodixand Content (g) 23066 2.2812 22689 2.2870 2.2710 2.2797 22621 22607 Adsat ent Type None CG161M CG71 M XAD-7 C930 Silica D350EL Silica XAD-2 XAD4 Adsorbent Mass (g) - 1.1438 1.1345 1.1511 1.1385 1.1422 1.1311 1.1402 Manufacturer - Rohm & Haas Rohm & Haas Rohm & Haas PQ Corp. PQ Corp. Rdvn &
Haas Rohm & Haas Lot# - 3558506 3604478 03729DF 52642 S009 0491717 03803HF
lodixand Mass/Solids Mass - 1.9944 2.0000 1.9868 1.9947 1.9959 2.0000 2.0003 Dryirg T C - 40 40 100 250 250 100 100 Equilibrium Analysis (Nommized %component in sdution) lodixand 99.2391 99.301 99.558 99.431 99.2569 99.3377 99.2592 99.214 Backpeaks 0.6829 0.665 0.4197 0.5205 0.6938 0.627 0.6535 0.7246 N ized 0.078 0.034 0.0222 0.0484 0.0493 0.0352 0.0873 0.0614 All patents, journal articles, publications and other documents discussed and/or cited above are hereby incorporated by reference.

Claims

We claim:

1. A process for purifying iodixanol from N-acetyl cyclized iodixanol, cyclized iodixanol, or primary o-alkyl trimer using non-polar organic adsorbents, where the average pore diameter is smaller than about 30 nm.