3/4/2011

1

Pharmaceutical Surveillance with RapidSpectroscopic Screening Technologies

Cindy BuhseDivision of Pharmaceutical Analysis (DPA), US FDA

Disclaimer:The findings and conclusions in this presentation have not been formally disseminated by the Food and Drug Administration and should not be construed to represent any Agency determination or policy.

1

Why Rapid Screening?• Rapid screening will dramatically increase the

number of containers of material that can benumber of containers of material that can be examined without a dramatic increase in personnel.– Materials failing rapid screening will be sent to

FDA laboratories for further testing.

• Rapid screening can support rapid response

2

• Rapid screening can support rapid response.

3/4/2011

2

Rapid Screening: Instruments and MethodsFour types of portable instruments have been evaluated for use at dockside or at manufacturing facilities and are currently being deployed. Materials failing

id i ld b ti d d t t FDA l b t i f f thrapid screening would be quarantined and sent to FDA laboratories for further testing :

•X-Ray Fluorescence (XRF) Spectrometer for detection of toxic metals

•Ion Mobility Spectrometer (IMS) for detection of weight loss drugs in dietary supplements

•Raman and Near Infra-Red (NIR) Spectrometers for detection of i i h i l i di

3

contaminants in pharmaceutical ingredients

Raman and NIR could also be used to identify or verify material if FDA had a robust/complete library.

Identification and Verification of Pharmaceutical Excipients with Spectral Libraries

• Acquire “certified” reference materials• Measure the spectra of reference materials and

store in a database for future use• Applications

4

– Material identification– Material verification

3/4/2011

3

The Hit Quality Index (HQI)

• Term derived from ID and verification applications• Term derived from ID and verification applications.

• The most common HQI is the spectral correlation coefficient.

HQI = 1.000 is best match, indicates perfect correlationHQI 0 is orst match indicates complete lack of

5

HQI = 0 is worst match, indicates complete lack of correlation

Typical Applications

• ID: material’s identity is unknown– Compare an unknown’s test spectrum to an entire

library of known spectra to determine its most likely identity.

• Library spectrum with highest HQI

• Verification: identity known but unconfirmedCompare a test spectrum to its library spectrum to

6

– Compare a test spectrum to its library spectrum to verify its identity

• Require the HQI to exceed a predetermined threshold

3/4/2011

4

Recent Work at the DPA

• Characterize the capabilities of spectral library-based correlation methods to identify and verify materials

• Evaluate procedures to transfer Raman and NIR libraries from one instrument to anotherD l d d d f t f i

7

• Develop advanced procedures for transferring Raman and NIR libraries.

Examples of Spectra and their HQIs: Master Instrument Spectra against Castor Oil

1 00

Na_CMC(HQI = 0.748)

Cocoa Butter (HQI = 0.840)

0.40

0.60

0.80

1.00

man

Inte

nsity

(nor

m.)

Sesame Oil (HQI = 0.847)

Castor Oil (Target, HQI=1.000)

8

0.00

0.20

200 700 1200 1700 2200Raman Shift (cm-1)

Ram

3/4/2011

5

Na_CMC(HQI = 0.00002)

Cocoa Butter (HQI = 0.229)

Derivative Spectra and their HQIs:Master Instrument Spectra against Castor Oil

0.10

0.15

0.20

man

Inte

nsity

(nor

m.)

Sesame Oil (HQI = 0.955)

Castor Oil (Target, HQI=1.000)

9

0.00

0.05

200 700 1200 1700 2200Raman Shift (cm-1)

Ram

Evaluation of HQI after Library TransferLambda Solutions P1 Raman (LSI)

Master library measured on benchtop instrument.

Ahura TruScanEnwave EZ-Raman

B&W T k Mi iR II

Delta Nu Reporter ID

p

Test samples measured on library instrument.

Test samples measured on four “field” instruments.

10

B&W Tek MiniRam II

3/4/2011

6

Benchmark: Test spectra measured on Master instrument without library transfer

VerificationAverage HQI = 0.98 ± 0.03

Sample Hit # HQI Sample Hit # HQI1 acetaminophen 1 0.989 19 ibuprofen 1 0.9792 acetylsalicylicacid 1 0.975 20 ketoprofen 1 0.9973 alphalactose 1 0.994 21 magnesiumstearate 1 0.9964 benzopril 1 0.985 22 melamine 1 0.9875 benzylalcohol 1 0.997 23 metformin 1 0.9936 caffeine 1 0.984 24 microcrystallinecellulose 1 0.9707 calciumphosphatedibasic 1 0.978 25 naproxensodium 1 0.9868 carbamezpene 1 0.994 26 PEglycol 1 0.9779 i i id 1 0 981 27 h 1 0 998

IDve age Q 0.98 0.03

11

9 citricacid 1 0.981 27 pgstarch 1 0.99810 crospovidone 3 0.980 28 propyleneglycol 1 0.99811 diclofenac 1 0.989 29 pseudoephedrine 1 0.97312 diethyleneglycol 1 0.999 30 sodiumcmcellulose 1 0.92413 estradiol 1 0.995 31 sorbitol 1 0.99814 ethyleneglycol 1 1.000 32 sulfanilamide 1 0.99015 gelatin 1 0.851 33 tinidazole 1 0.99316 glycerin 1 0.999 34 triclosan 1 0.97517 HPMC 1 0.964 35 xylometazoline 1 0.99318 hydrocortisone 1 0.978

Library Transfer Across Instruments:Uncorrected First Derivative Spectra

IDVerification

0.982

0.459

0.785

0.495 0.491

0.200

0.400

0.600

0.800

1.000

Ave

rage

HQ

I

100 94.3 100 100 100

20

40

60

80

100

% C

orre

ct ID

12

0.000LSI Enwave Ahura BWTek DeltaNu

0LSI Enwave Ahura BWTek DeltaNu

3/4/2011

7

Library Transfer Across Instruments:Corrected First Derivative Spectra

IDVerification100 100 97.1 100 100

20

40

60

80

100

% C

orre

ct ID

0.9780.912 0.937 0.910

0.830

0.300

0.500

0.700

0.900

1.100

Ave

rage

HQ

I

13

0LSI Enwave Ahura BWTek DeltaNu-0.100

0.100

LSI Enwave Ahura BWTek DeltaNu

Shift correctionIntensity correctionResolution correction

Recent Work at DPA

• Determine the sensitivity of Raman library-based spectral correlation methods to the presence of contaminants in excipients– Economically motivated adulterants (EMAs)

• Develop advanced procedures to improve the sensitivity to contaminants in excipients

14

sensitivity to contaminants in excipients– Chemometrics

3/4/2011

8

Raman Spectra

Contaminants

Ram

an In

tens

ity

Propylene Glycol

Diethylene Glycol

Ethylene Glycol

Excipients

15

250 750 1250 1750 2250 2750

Raman Shift (cm-1)

Glycerin

Propylene Glycol

Sensitivity to Contaminants: One Instrument

% DEG0.0010.0020.0030.0040.0050.0060.0070.0080.0090.00100.00

HQI Threshold 0.950

% DEG0.0010.0020.0030.0040.0050.0060.0070.0080.0090.00100.00

% DEG0.0010.0020.0030.0040.0050.0060.0070.0080.0090.00100.00

HQI Threshold 0.950

Qua

lity

Inde

x (H

QI) H

it Quality Index (H

0.400

0.500

0.600

0.700

0.800

0.900

1.000

0.400

0.500

0.600

0.700

0.800

0.900

1.000

Glycerin

DEG

~ 18%

Qua

lity

Inde

x (H

QI) H

it Quality Index (H

0.400

0.500

0.600

0.700

0.800

0.900

1.000

0.400

0.500

0.600

0.700

0.800

0.900

1.000

Glycerin

DEG

Qua

lity

Inde

x (H

QI) H

it Quality Index (H

0.400

0.500

0.600

0.700

0.800

0.900

1.000

0.400

0.500

0.600

0.700

0.800

0.900

1.000

Glycerin

DEG

Glycerin

DEG

~ 18%

16

% Glycerin

Hit

HQ

I)

0.000

0.100

0.200

0.300

0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.000.000

0.100

0.200

0.300

% Glycerin

Hit

HQ

I)

0.000

0.100

0.200

0.300

0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.000.000

0.100

0.200

0.300

% Glycerin

Hit

HQ

I)

0.000

0.100

0.200

0.300

0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.000.000

0.100

0.200

0.300

3/4/2011

9

80

Determination of DEG in glycerin by partial least squares regression is very sensitive.

y = 0.99x + 0.1R2 = 0.9999

40

60

ed D

EG

Com

posi

tion

(%)

0.2%

17

0

20

0 10 20 30 40 50 60 70 80

Measured DEG Composition (%)

Pred

icte

Conclusions

• Baseline removal is essential for HQI sensitivity.• Corrections have a significant impact on HQI values and are• Corrections have a significant impact on HQI values, and are

thus essential for verification.– Shift correction-most sensitive– Intensity correction– Resolution correction-when bandwidth differs from library

• ID analysis of the major component is not sensitive to spectral corrections.– ID is observed to be highly accurate for major component, but not for

18

g y j p ,low concentration components/impurities.

• Detection of low levels of contaminants/impurities may require chemometric model development, but is still possible by spectroscopic methods.

3/4/2011

10

Metal/Element work at DPA

• Developed X-ray fluorescence (XRF) to measure metal impurities in drug substances and productsmetal impurities in drug substances and products– Simple, fast, no sample prep, inexpensive

• Determine if XRF is capable of quantifying residual catalysts at the levels specified by the EMA Guideline (expected ICH guideline)– Spiked excipients– Measure against ICP-MS reference method

19

g• Determine if metal/element profiling in excipients can

predict the source– Ca, Mg, S, Cl, Br, Fe, Cu, Ti, Si, etc.– Verify XRF with ICP-MS

X-ray Fluorescence• Elemental impurities in pharmaceutical materials• Toxics• Toxics

– As, Pb, Hg, Cr– LoDs 8, 14, 20 and 150 ppm

• Catalysts– LoDs in the 50-75 ppm range

20

pp g

3/4/2011

11

3000 250

Example: 25 ppm Arsenic in solution (green), powder (red) and tablet (black)

Cou

nts

1000

1500

2000

2500

10000 120000

50

100

150

200

21

Energy / eV5000 10000 15000 20000 25000 30000

0

500

Role of IPEC

• Provide materials to create library and study metals/elements in excipients– We would like to have qualified excipients from

several vendors– Preferably several lots from each vendor in order

to establish normal variability in spectral

22

o es b s o v b y specproperties

• Expertise on which excipients to target

3/4/2011

12

How

• Material Transfer Agreement with Division of• Material Transfer Agreement with Division of Pharmaceutical Analysis– To allow for DPA to accept material– To allow for DPA to share results with IPEC

members (if desired)

23

Most wanted excipientsBUTYLATED HYDROXYANISOLE

GELATIN POVIDONE

BUTYLATED HYDROXYTOLUENE

Glycerin PREGELATINIZED STARCH HYDROXYTOLUENE

CALCIUM PHOSPHATE (DIBASIC)

HYDROXY PROPYL CELLULOSE

Propylene Glycol (liquid and solid)

CALCIUM STEARATE HYDROXY PROPYL METHYLCELLULOSE

SHELLAC

CARBOXYMETHYL CELLULOSE

LACTOSE SILICON DIOXIDE

CASTOR OIL MAGNESIUM STEARATE SODIUM STARCH GLYCOLATE

24

CRESOL MALTODEXTRIN Sorbitol

CROSCARMELLOSE MICROCRYSTALLINE CELLULOSE

STARCH (CORN)

CROSPOVIDONE Polyethylene Glycol, liquid and solid

STEARIC ACID

ETHYLCELLULOSE POLYSORBATE 80 Talc

Titanium Dioxide

3/4/2011

13

Questions?

25

References:

• Pharmaceutical Surveillance with Rapid Spectroscopic Screening i ff l A i h i l iTechnologies JF Kauffman et al, American Pharmaceutical Review,

2010• Multivariate Calibration Standardization Transfer Across Multiple

Instruments for the Rapid Detection of Diethylene Glycol in Glycerin by Raman Spectroscopy CM Gryniewicz-Ruzicka et al, Applied Spectroscopy 2011

• Using Portable Ion Mobility Spectrometer to Screen Dietary Supplements for Sibutramine JD Dunn et al, Journal of Pharmaceutical and Biomedical Analysis, 2011

26

y• Detection of Diethylene Glycol Adulteration in Propylene Glycol –

Method Validation through a Multi-Instrument Collaborative Study X Li et al, Journal of Pharmaceutical and Biomedical Analysis, 2011

• Rapid Limit Tests for Metal Impurities in Pharmaceutical Materials by X-ray Fluorescence Spectroscopy using the Continuous Wavelet Transform S Arzhantsev et al, Analytical Chemistry, 2011