BAFFI ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 4, 2000 793 DRUGS, COSMETICS, FORENSIC SCIENCES

Quantitative Determination of Diethylene Glycol Contamination in Pharmaceutical Products PATRICIA BAFFI and SURIA ELNESER Universidad Central de Venezuela, Facultad de Farmacia, Apdo 40109, Caracas 1040-A, Venezuela MYRNA BAFFI and MATILDE DE MELIN Instituto Nacional de Higiene “Rafael Rangel,” Departamento de Quimica de Medicamentos Ciudad Universitaria (UCV) Apdo 60412, Oficina del Este, Caracas, Venezuela

A simple, rapid, and reliable method was developed for the quantitative determination of diethylene glycol (DEG) in pharmaceutical products using capillary gas chromatography with flame ionization detection. The method uses ethylene glycol as internal standard and allows for the separation of propylene glycol and DEG. The assay was linear in a DEG concentration range between 1.0 and 10.00 mg/mL, with coefficients of variation of 2.3–4.4% for the tested concentrations. Quantitation and detection limits, respectively, were 1.0 mg/mL and 0.15 mg/mL diethylene glycol. The method was used to analyze 3 pharmaceutical products possibly contaminated with diethylene glycol, of which one was suspected of causing intoxication and death in children. Infrared spectroscopy was used to confirm the identity of diethylene glycol. This analytical methodology is proposed for evaluation of pharmaceutical products containing glycols to prevent intoxication and for security level verification.

ropylene glycol (1,2-propanediol), ethylene glycol (1,2-ethanediol), and diethylene glycol (bis [2-hydroxyethyl] ether), are chemically classified as glycols. Propylene glycol is a frequent vehicle and solvent in oral, topical, and parenteral pharmaceutical products. A U.S. Pharmacopeia (USP) monograph describes propylene glycol as a solvent with a number of acceptable pharmaceutical uses, and it is classified by the U.S. Food and Drug Administration (FDA) as generally recognized as safe for appropriate food use (1). The maximum daily intake of propylene glycol for humans, as calculated by the FDA (25 mg/kg), does not include amounts absorbed from medications. Although it is considered to be the least toxic of the glycols, several incidents of toxicity in children may have been caused by administration of medications containing multiple doses of propylene glycol (2, 3). Ethylene glycol and diethylene glycol are widely used in chemical industries as solvents or antifreeze products that may

P

Received July 27, 1999. Accepted by JM January 21, 2000.

cause neuro- and nephrotoxicity in humans (4, 5). Because there is no USP monograph on ethylene glycol or diethylene glycol, their application in a pharmaceutical formulation would violate good manufacturing practices. The USP monograph on polyethylene glycol, which is used in oral pharmaceutical products, establishes limits of ethylene glycol and diethylene glycol at not more than 0.25%, combined total. It is obvious, therefore, that the use of both diethylene glycol and ethylene glycol as vehicles for oral pharmaceuticals is not allowed. Poisoning associated with diethylene glycol ingestion through manufactured pharmaceutical products has been reported by the World Health Organization (WHO). Incidents in the United States, Nigeria, and Haiti (6–9) were due to inadequate control of raw materials and final products. In Maracaibo, Venezuela (1994), the ingestion of a children’s syrup containing acetaminophen and glycol resulted in intoxication and death. Samples of the suspected products taken in Maracaibo and other regions of the country were sent to the Instituto Nacional de Higiene “Rafael Rangel” for analysis to determine if diethylene glycol was present in the acetaminophen syrup and in 2 other pharmaceutical products from the same laboratory: one containing vitamin B complex and the other, ferrous sulfate and vitamin B12. This study describes a 3-step analytical methodology for the qualitative and quantitative determination of the toxic component: (1) identification of propylene glycol, ethylene glycol, and diethylene glycol; (2) quantitative determination of diethylene glycol; and (3) confirmation of diethylene glycol. Capillary gas chromatography (GC) was used for steps 1 and 2, and infrared (IR) spectroscopy of a previous thin layer chromatography (TLC) separation was used for step 3. The capillary GC analysis was validated in terms of accuracy, precision, linearity, sensitivity, selectivity, and statistical data treatment. Capillary GC has been widely used for diethylene glycol determination in wines, manufactured food products, and human sera (10–12). Most of these studies involved trimethylsilyl derivatives of diethylene glycol, glycol extraction from aqueous solutions, and evaporation, resulting in low recoveries (13). Serum ethylene glycol determination in the presence of other glycols such as propylene glycol, 1,3-butanediol, and diethylene glycol, has been previously re-

794 BAFFI ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 4, 2000

ported without derivatization and under isothermal conditions (14). Glycol analysis must be performed at low temperature to avoid thermal decomposition. If derivatization is used, the possible formation of other products or incomplete reaction must be considered (15). Therefore, we determined diethylene glycol without derivatization and under low temperature isothermal conditions. Capillary GC combined with mass spectrometry (GC/MS) is the best choice for glycol confirmation (15, 16). In the case of laboratories that are not equipped for this technique, TLC (17) followed by sample extraction and IR spectroscopy (18) is a satisfactory alternative. Kenyon et al. (19) determined diethylene glycol in glycerin and elixirs by TLC and densitometry. The Centers for Disease Control and Prevention (CDC), FDA, and WHO established a detection level of 0.1% diethylene glycol for screening raw material and glycerin-based elixirs.

Instrument conditions were nitrogen carrier gas flow rate at 3.5 mL/min, hydrogen flow rate at 30.0 mL/min, air flow rate at 300.0 mL/min, injector temperature 260ºC, detector temperature 300ºC, column temperature 100ºC, attenuation 4 and range 1, injection volume 1.0 µL and 10:1 split.

Standards of Preparation Stock standards were prepared by separately diluting each standard in distilled water at a concentration of 0.05 g/mL diethylene glycol and 0.01 g/mL each of ethylene glycol and propylene glycol. For glycol identification, each standard solution was prepared in distilled water at a concentration of 2.5 mg/mL diethylene glycol and 1.0 mg/mL each of ethylene glycol and propylene glycol. Diethylene glycol concentration in the standard solutions for the calibration curve were 1.0, 2.5, 5.0, 7.5, and 10.0 mg/mL with 1.0 mg/mL ethylene glycol as internal standard.

Experimental

Preparation of Samples Materials and Reagents All reagents were of analytical grade: acetone, sulfuric acid, ammonium hydroxide (Riedel de Haen, A.G, Seelze, Germany), chloroform, and ethanol (E. Merck, Darmstadt, Germany). Sulfuric potassium dichromate solution was prepared by dissolving 5.0 g potassium dichromate in 40% 100.0 mL sulfuric acid solution. Deionized water was obtained from a Milli Q plus system. Propylene glycol (BDH Chemicals Ltd., Poole, UK), ethylene glycol (E. Merck), and diethylene glycol (Chem Service, West Chester, PA) >99% pure were used as standards. All samples tested were commercial syrups obtained from the laboratory that produced the suspected children’s syrup. (a) Antipyretic syrup.—Five samples of antipyretic syrup (A–E), declaring n-acetyl-p-aminophenol, 120 mg; sodium saccharin, 40 mg; propylene glycol, 1.4 g; glycerin, 1.1 g; 70% sorbitol, 2.2 g; and distilled water to 5.0 mL. (b) Polyvitaminic syrup.—Four samples of polyvitaminic syrup (A–D) declaring vitamin B1, 80 mg; vitamin B2, 40 mg; vitamin B6, 40 mg; vitamin B12, 60 mg; niacinamide, 280 mg; sodium saccharin, 0.12 g; and distilled water to 100.0 mL. (c) Antianemic syrup.—Four samples of antianemic syrup (A–D) declaring ferrous sulfate, 3.33 g; vitamin B12, 20 mg; sodium saccharin, 0.12 g; vehicle (glycerin, sugar, and water) to 100.0 mL.

Capillary Gas Chromatograph Capillary GC analysis was performed on a Perkin-Elmer (Norwalk, CT) Autosystem gas chromatograph equipped with an automatic sampler and injection system, split injector with injection liner packed with glass wool (4 mm id and 6 mm od), and flame ionization detector. Chromatography column SPB-1 fused silica (Supelco, Inc., Bellefonte, PA) 15 m × 0.53 mm id, 1.5 µm film thickness, was used. Chromatograms were recorded and integrated with a PE Nelson Model 1020 integrator and an Epson LX-810L printer.

For glycol identification purposes, assay syrups were diluted as follows: (a) Antipyretic syrup.—1.0 mL syrup diluted to 25.0 mL with distilled water in a volumetric flask (solution A); 3.0 mL Solution A was then diluted to 10.0 mL with distilled water in a volumetric flask. (b) Polyvitaminic syrup.—1.0 mL syrup diluted to 10.0 mL with distilled water in a volumetric flask. (c) Antianemic syrup.—1.0 mL syrup diluted to 25.0 mL with distilled water in a volumetric flask (solution A); 4.0 mL Solution A was then diluted to 10.0 mL with distilled water in a volumetric flask. For quantitative determination of diethylene glycol, samples were treated in the same way as for glycol identification, with 1.0 mg/mL internal standard added to the final solution.

Diethylene Glycol Calibration Curve and Precision Studies The calibration curve was obtained by plotting the diethylene glycol concentration of each standard solution against the peak area ratios of diethylene glycol and internal standard. The linear regression was obtained from 5 determinations of each standard concentration. The precision study was reported as the coefficient of variation (CV) for n = 5. Concentration of diethylene glycol samples was determined by interpolation in the calibration curve with the dilution factor taken into account. Precision studies obtained from sample analysis were also reported as the CV (n = 5).

Statistical Data Treatment The statistical analysis (20) includes confidence limits of 95 and 99% for diethylene glycol determination in syrups, variance analysis (ANOVA), and Duncan test for the means comparison.

BAFFI ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 4, 2000 795

Figure 1. Chromatographic separation of standards: ethylene glycol, propylene glycol, and diethylene glycol.

Thin-Layer Chromatography and Infrared Spectroscopy IR spectroscopy was used for diethylene glycol confirmation in syrup, after TLC identification and separation of glycol and glycerol (17). Equipment.—TLC was performed in a 30 × 15 cm chromatographic tank. Chromatographic plates of Silica gel 60 without fluorescent indicator, 20 × 20 cm and 0.25 mm wide (Merck) were used. The mobile phase consisted of a mixture of chloroform, acetone, and 5N NaOH (10 + 80 + 10) with previous saturation of the chromatographic chamber for 1 h. A Perkin-Elmer 1310 IR spectrophotometer was used to obtain the IR spectrum in the range of 4000 and 400 cm–1. A mixture of 1.0 mL diethylene glycol and 2.0 mL ethanol was used as standard preparation for Rf determination. The sample solution was a mixture of 1.0 mL syrup and 1.0 mL ethanol. Procedure.—A 5.0 µL amount of both standard and sample solutions was placed in 2 separate chromatoplates. After separation, one of the plates was sprayed with potassium dichromate-sulfuric acid. The high colored spots were delimited and Rf value of diethylene glycol was determined

(Rf = 0.46) as a guideline to recognize the glycol position zones in the other plate. Silica gel zones were scraped using the Rf determined from the delimited plate and transferred to centrifuge tubes. Chloroform (10.0 mL) was added to each tube, which was then agitated for 5 min, and centrifuged. A 7.0 mL aliquot of supernatant liquid (sample and standard, respectively) was transferred to a 10.0 mL beaker and evaporated. Finally, the IR spectrum of the residue was scanned between 4000 and 400 cm–1, and a film was placed in a sodium chloride crystal for characteristic band absorption identification purposes.

Table 1. Total retention times of glycols (tR) Glycols

tRa

Ethylene glycol

1.708

Propylene glycol

1.887

Diethylene glycol

4.048

a

Means of 5 determinations (min).

796 BAFFI ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 4, 2000

Figure 2. Chromatograms obtained in analysis of diethylene glycol in samples (– – –), compared with standards (———). (A) Antipyretic syrup.

Figure 2. Chromatograms obtained in analysis of diethylene glycol in samples (– – –), compared with standards (———). (B) Polyvitaminic syrup.

BAFFI ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 4, 2000 797

Figure 2. Chromatograms obtained in analysis of diethylene glycol in samples (– – –), compared with standards (———). (C) Antianemic syrup.

Results and Discussion A capillary GC method was developed for determination of diethylene glycol in pharmaceutical products and formulations containing propylene glycol. The method is simple and rapid and does not require previous sample cleanup or glycol derivatization. The assay is carried out at a low isothermal temperature and gives good separation between propylene, diethylene, and ethylene glycols. The calculated ethylene glycol and propylene glycol resolution was 2.3, which is better than pre-

Table 2. Precision studies of the calibration curve for diethylene glycol (DEG) by the internal standard (IS) method (n = 5) Concentration, mg/mL

Area, DEG/IS

CV, %

1.0

0.878761

2.31

2.5

2.430658

2.66

5.0

4.856771

2.74

7.5

7.186512

4.38

9.413932

2.77

10

viously reported (resolution = 1.5; 14). A resolution >1.5 is usually considered to be a complete separation (99.7%; 16). Figure 1 shows the chromatogram of the standard mixture (ethylene glycol, propylene glycol, and diethylene glycol) and the total retention time of each separated compound as presented in Table 1. Sample chromatograms were compared with those of the standard glycol mixture (Figure 2). Diethylene glycol was present in samples A, B, and C, and propylene glycol was found in the antianemic syrup C. Ethylene glycol was not found in any of the samples. The calibration curve for diethylene glycol using the internal standard method was linear within the concentration range of 1.0–10.0 mg/mL. Equation of the line regression was y = 0.947 x + 0.027; r = 0.9997; S (y/x) = 0.099. The standard deviation of the slope was 0.013 and that of the intercept 0.083. The 95% confidence limit levels for the slope were 0.947 ± 0.041 and 0.027 ± 0.265 for the intercept. Diethylene glycol detection and quantitation limits were 0.30 mg/mL and 1.0 mg/mL, respectively, calculated from the intercept and S (y/x; 20). The experimental value for the detection limit was 0.15 mg/mL with a signal to noise ratio of 8:1 and a CV of 14.65% (n = 5). The relatively high CV may be attributable to the fact that the internal standard calculation method was not used to determine the detection limit. The quantitation limit experimentally obtained was 1.0 mg/mL.

798 BAFFI ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 4, 2000

The precision results for the diethylene glycol standard concentration in the calibration curve show a CV 2.31–4.38% (Table 2). For the diethylene glycol samples analysis using internal standard, the CV ranged between 0.91 and 4.89% (Table 3). The diethylene glycol concentration in the samples ranged from 27.14 to 28.98% (w/v) in the antipyretic syrups, from 4.05 to 4.09% (w/v) in the polyvitaminic syrups, and from 23.54 to 25.72% (w/v) in the antianemic syrups. Mean confidence limits at 95 and 99% for several samples are shown in Table 4. The diethylene glycol mean values were 27.73, 4.06, and 24.78% (w/v) for the antipyretic, polyvitaminic, and antianemic syrups, respectively. The accuracy of the results ranged from 96.19 to 104.51% for the antipyretic syrup, 99.67 to 100.83% for the polyvitaminic syrup, and 94.98 to 103.77% for the antianemic syrup. CVs were 3.18, 0.69, and 4.38%, respectively. The diethylene glycol concentration in all samples was within a mean confidence limit of 95 and 99%, except for syrup A, which did not fulfill the 95% requirement. Table 5 lists the variance analysis and Duncan test for comparison of means. The antipyretic syrup showed 2 homogeneous groups of p < 0.01 and 3 homogeneous groups of p < 0.05. The polyvitaminic syrup did not show significant difference between samples A, B, C, and D, and the antianemic syrup showed 2 homogeneous groups of p < 0.01 and p < 0.05.

Table 3. Results obtained for diethylene glycol analysis of samples by the internal standard method (n = 5) Samples

Area

CV, %

Concentration, % (w/v)

Antipyretic A

3.322189

2.12

28.98

B

3.060019

2.51

26.68

C

3.207725

4.89

27.98

D

3.19798

3.19

27.89

E

3.112137

2.72

27.14

Polyvitaminic A

3.864227

3.29

4.05

B

3.848044

2.36

4.03

C

3.908905

2.83

4.09

D

3.88604

0.91

4.07

Antianemic A

3.594826

2.15

23.54

B

3.697515

3.73

24.21

C

3.925348

2.69

25.72

D

3.917567

3.65

25.67

Table 4. Accuracy and precision results for diethylene glycol analysis in samples Confidence limit of means Samples

Obtained value, % (w/v)

Mean value, % (w/v)

Accuracy, %

CV, %

p = 0.01

p = 0.05

3.18

27.73 ± 1.818

27.73 ± 1.096

0.69

4.06 ± 0.082

4.06 ± 0.044

4.38

24.78 ± 3.168

24.78 ± 1.726

Antipyretic A

28.98

27.73

104.51

B

26.68

27.73

96.19

C

27.98

27.73

100.88

D

27.89

27.73

100.57

E

27.14

27.73

97.85 Polyvitaminic

A

4.05

4.06

99.67

B

4.03

4.06

99.25

C

4.09

4.06

100.83

D

4.07

4.06

100.24 Antianemic

A

23.54

24.78

94.98

B

24.21

24.78

97.70

C

25.72

24.78

103.77

D

25.67

24.78

103.56

BAFFI ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 4, 2000 799 Table 5. Statistical analysis for comparison of means Variance analysis

Means comparisons (Duncan’s test) Homogeneous groups

Source

DF

SS

MS

F

Variable

Mean

p < 0.01b

p < 0.05c

Antipyretic syrups A, B, C, D, E

Between

4

15.5974

3.89935

Within

20

16.4274

0.82137

Total

24

32.0248

4.75a

A

28.984

I

I

C

27.977

II

II

D

27.892

II

III

E

27.136

I

II

B

26.678

I

I

C

4.0975

I

I

D

4.0733

I

I

A

4.0503

I

I

B

4.0332

I

I

p < 0.01e

p < 0.05f

Polyvitaminic syrups A, B, C, D

Between

3

0.01171

0.0039

Within

16

0.16858

0.01054

Total

19

0.18029

0.37 (NS)d

Antianemic syrups A, B, C, D

Between

3

17.6483

5.88276 0.61703

Within

16

9.8725

Total

19

27.5208

9.53 (S)a

C

25.718

I

I

D

25.666

I

I

B

24.215

II

I

A

23.537

I

I

a

Significant, p