ATR-FTIR characterization of generic brand-named and counterfeit sildenafil- and tadalafil-based tablets found on the Brazilian market

    ATR-FTIR characterization of, generic, brand-named and counterfeit sildenafil- and tadalafil-based tablets found on the Brazilian market Jos´e Coelho Neto, Fernanda L.C. Lisboa PII: DOI: Reference:

S1355-0306(17)30055-2 doi: 10.1016/j.scijus.2017.04.009 SCIJUS 665

To appear in:

Science & Justice

Received date: Revised date: Accepted date:

24 January 2017 3 April 2017 20 April 2017

Please cite this article as: Jos´e Coelho Neto, Fernanda L.C. Lisboa, ATR-FTIR characterization of, generic, brand-named and counterfeit sildenafil- and tadalafil-based tablets found on the Brazilian market, Science & Justice (2017), doi: 10.1016/j.scijus.2017.04.009

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

SC

RI

PT

ATR-FTIR characterization of, generic, brand-named and counterfeit sildenafil- and tadalafil-based tablets found on the Brazilian market

Abstract

NU

Viagra and Cialis are among the most counterfeited medicines in many parts of the world, including Brazil. Despite many studies have been made regarding discrimina-

MA

tion between genuine and counterfeit samples, most published works do not contemplate generic and similar versions of these medicines and also do not explore excipi-

D

ents/adjuvants contributions when characterizing genuine and suspected samples. In this

TE

study, we present our findings in exploring ATR-FTIR spectral profiles for characterizing both genuine and questioned samples of several generic and brand-name sildenafil-

AC CE P

and taladafil-based tablets available on the Brazilian market, including Viagra and Cialis. Multi-component spectral matching (deconvolution), objective visual comparison and correlation tests were used during analysis. Besides from allowing simple and quick identification of counterfeits, results obtained evidenced the strong spectral similarities between generic and brand-named tablets employing the same active ingredient and the indistinguishability between samples produced by the same manufacturer, generic or not. For all sildenafil-based and some tadalafil-based tablets tested, differentiation between samples from different manufacturers, attributed to slight variations in excipients/adjuvants proportions, was achieved, thus allowing the possibility of tracing an unknown/unidentified tablet back to a specific manufacturer. Keywords: Viagra, Cialis, Excipients, Generic medicine, Manufacturer identification,

Preprint submitted to Science & Justice

April 20, 2017

ACCEPTED MANUSCRIPT

Spectral deconvolution.

PT

1. Introduction

RI

It is nearly 20 years now since Viagra, the first phosphodiesterase 5 (PDE5) inhibitor, reached the market and transformed erectile disfunction (ED) treatment[1, 2]. Aside

SC

from its initial development as a potential treatment for angina and its enormous success as an on-demand oral treatment for ED, today’s therapeutic potencial uses of sildenafil

NU

and other PDE5 inhibitors, like tadalafil, include treatment of pulmonary hypertension, heart disease, diabetes, cancer and benign prostatic hyperplasia, among other possible

MA

indications[1, 3, 4]. Even if new uses for PDE5 inhibitors were to be disregarded, the estimated worldwide increase of ED cases, possibly afflicting more than 322 million men

D

around 2025[5], coupled with recreational use and misuse, in spite of the potencial phys-

TE

ical and psychological dangers[6–8], should be more than enough to keep the ongoing demand (prescribed or not) for this drug class.

AC CE P

As a side-effect of its own success, PDE5 inhibitors are among the most counterfeited medicines around the world, specially in Brazil, where about two thirds of counterfeit medicines seized by local authorities belong to this class[9, 10]. From the forensic point of view, analysis of questioned PDE5 samples and identification of counterfeits has been achieved based on multiple approaches, including physical profiling(PP)[11, 12], Ultra Performance Liquid Chromatography with Electro Spray Ionization and Mass Spectrometry(HPLC-ESI-MS)[13], Near Infrared (NIR) and Raman Spectrometry[14, 15] and Attenuated Total Reflectance Fourier Transform Infrared Spectrometry (ATR-FTIR)[16]. The great majority of published works, however, is focused on Viagra and Cialis alone and only considers the infrared spectra of sildenafil and tadalafil, mostly ignoring that each tablet is a solid mixture composed by the active ingredient (AI) and several other 2

ACCEPTED MANUSCRIPT

excipients/adjuvants. In fact, for Viagra (50mg) and Cialis (20mg), excipients/adjuvants mass amount to 77.7% and 94.5% of each tablet, respectively. Therefore, their presence

PT

and contributions to the spectral profile of the tablets must be taken into account during

RI

analysis.

Also, the ongoing demand for PDE5 inhibitors, coupled with patent expirations and the

SC

expansion of generic drugs market in many countries, including Brazil, prompted many generic and brand-named alternative versions of Viagra and Cialis to be available at drug-

NU

stores. Many of these new sildenafil- and tadalafil-based tablets are manufactured by companies other than Pfizer (maker of Viagra) and Eli Lilly (maker of Cialis), and, therefore,

MA

may present different spectral characteristics, regardless of bioequivalency. As counterfeit versions of these new tablets may soon be reaching the streets, new studies concerning

D

their characteristics, specially when compared to Viagra and Cialis, are needed.

TE

Here we report our findings in characterizing some of the sildenafil- and tadadafilbased tablets legally available on the Brazilian market. Samples studied encompass not

AC CE P

only Viagra and Cialis, but several other tablets, generic and brand-named, manufactured and marketed by different pharmaceutical companies. Analysis of ATR-FTIR spectral profiles was used to identify not only the AI of each sample, but also the main excipients/adjuvants present. Multi-component spectral analysis (deconvolution), objective visual comparison and correlation allowed genuine samples from different manufacturers to be distinguished from each other. The same procedures were successfully used to identify unknown and counterfeit seized samples. All analyses were made without explicitly recurring to chemometrics, the technique usually employed on most previous published works, thus making the proposed approach more easily accessible to forensic analysts working at laboratories equipped with modern ATR-FTIR spectrometers.

3

ACCEPTED MANUSCRIPT

2. Materials and Methods

PT

2.1. Samples Genuine tablets of sildenafil 50mg and tadalafil 20mg were purchased from local and

RI

regional trusted drugstores. Sildenafil samples consisted of 26 generic tablets from 7 phar-

SC

maceutical companies and 19 branded tablets from 6 pharmaceutical companies. Tadalafil samples consisted of 13 generic tablets from 5 pharmaceutical companies and 10 branded

NU

tablets from 4 pharmaceutical companies. Viagra and Cialis samples were included as part of the brand-named tablets. The brand name, pharmaceutical company, manufacturer and

MA

composition of each type of tablet were obtained from the packages and accompanying leaflets. The mass, shape and color of each individual tablet were also registered. Questioned tablets, seized by local law enforcement forces, consisted of 24 tablets

D

packaged as generic sildenafil from 2 pharmaceutical companies, 13 tablets packaged as

TE

Viagra, 1 tablet packaged as Dejav´u and 12 tablets packaged as Cialis. Other 15 unpackaged/unbranded tablets with unrecognizable or no identification marks, but suspected to

AC CE P

be PDE5 inhibitors, were treated as questioned tablets as well. After external physical characteristics (mass, shape and color) were registered, the external layer of each tablet was carefully removed. The remaining cores were then crushed and homogenized.

The total number of tablets studied was 133 (45 genuine sildenafil-based, 23 genuine tadalafil-based, 38 questioned/seized sildenafil-based, 12 questioned/seized tadalafil-based and 15 questioned/seized totally unknown tablets). 2.2. Instrumentation ATR-FTIR spectra of the crushed tablet cores were taken using a NicoletT M iZ10 spectrometer equipped with EverGlo IR source, DLaTGS room temperature IR detector and

4

ACCEPTED MANUSCRIPT

single-bounce Smart OrbitT M accessory module with diamond ATR crystal. All hardware from Thermo Fischer Scientific Inc. (USA).

PT

Each spectrum was averaged over 16 scans, taken at 4cm−1 resolution, maximum de-

RI

tector window aperture and minimum interferometer mirror speed, in the range of 4000 −

SC

400cm−1 . Background signal was averaged over 8 scans prior to each measurement. 2.3. Data analysis

NU

Analysis of collected spectra was carried out using the software suite accompanying the spectrometer, which included the OMNICT M 9.1.27, used for data acquisition and sin-

MA

gle spectrum comparisons (including correlation), and the OMNICT M SpectaT M 2.0, used for multi-component database comparison (deconvolution). The Specta software allows for multi-component spectral matching of mixtures containing up to 4 unknown compo-

D

nents or 1 known component (spectral bulk) and 4 unknown contaminants. All software

TE

from Thermo Fischer Scientific Inc. (USA).

AC CE P

A spectral IR library, containing profiles of pharmaceutical-grade raw materials purchased from local compounding pharmacies, was previously build and used as reference database for multi-component/deconvolution spectral analysis, as well as for the objective visual comparison between tested tablets. The library included reference spectra for both AI studied (sildenafil and tadalafil ) and most excipients/adjuvants listed by manufacturers in the leaflets accompanying genuine tablets (microcrystalline cellulose, lactose, magnesium stearate, calcium phosphate dibasic and sodium croscarmellose), along with more than 100 other ingredients employed by pharmaceutical companies (AIs and excipients/adjuvants), most of them found on tablet formulations. For the characterization and comparison between genuine samples, the spectral profiles of all individual tablets sharing the same identification, regardless of lot/batch numbers, were averaged altogether, thus producing a single spectral profile for each type of genuine 5

ACCEPTED MANUSCRIPT

tablet studied. The averaged profiles were normalized and used as a reference standard for the corresponding type of tablet. Each standard thus generated was then submitted to

PT

multi-component spectral matching (deconvolution), in order to assess which constituents

RI

could be automatically identified, allowing imediate confrontation with leaflet data. Correct identification of the tablets core composition was enhanced through the use of the

SC

spectral IR library generated from the same experimental setup and containing reference spectra for all target compounds. Objective visual comparison between reference profiles

NU

for each type of tablet was performed by overlaying all profiles sharing the same AI and manually checking for similarities and differences, taking into account specific spectral

MA

bands from each component previously identified, specially the excipients/adjuvants. In the final stage of the characterization/comparison, the correlation between all reference

D

profiles sharing the same AI was measured.

TE

Analysis of questioned tablets was conducted much the same way as the characterization and comparison between genuine tablets, whenever possible, with minor modifi-

AC CE P

cations. Seized tablets packaged together had their mass and color evaluated prior to the removal of the external layer and tablet core crushing and homogenization. A single average profile was produced from all tablets within the same package. The averaged profile from each questioned package was then compared to the standard corresponding to the package identification, the same occurring for tablet mass and color. For unpackaged (unmarked/unindentified) questioned tablets, some of which were received broken or even partially crushed, evaluation of mass and color could not be properly made and were thus disregarded. The spectral profile of each individual unidentified sample was normalized and compared to each of the genuine standards available for composition, spectral profile and correlation, in order to verify if a reasonable match could be found.

6

ACCEPTED MANUSCRIPT

3. Results and Discussion

PT

3.1. Genuine Tablets (generic and brand-named) 3.1.1. Sildenafil-based

RI

The nominal and physical characteristics obtained for the genuine sildenafil-based

SC

samples studied were resumed on Table 1.

Table 1: Nominal and physical characteristics of all genuine 50mg sildenafil-based tablets studied. The mass

Company

Manufacturer

Mass (g)

Shape

Color

Group

Viagra

Pfizer

Pfizer

3

0.314

diamond

blue

A1

Viasil

Teuto

Pfizer

2

0.321

round

blue

A2

Generic S1

Teuto

Pfizer

4

0.316

round

blue

A3

Dejav´u

Eurofarma

Eurofarma

6

0.324

pentagon

blue

B1

Generic S2

Eurofarma

Eurofarma

4

0.323

pentagon

blue

B2

Generic S3

Althaia

Eurofarma

2

0.322

pentagon

blue

B3

Vir´ıneo

Neo Qu´ımica

Brainfarma

2

0.313

oblong

blue

C1

AC CE P

TE

MA

Samples

D

Brand

NU

column contains the average mass per tablet of each brand.

Generic S4

Neo Qu´ımica

Brainfarma

8

0.321

oblong

blue

C2

Ah-Zul

Legrand

EMS

2

0.367

round

blue

D1

Generic S5

Germed

EMS

2

0.363

round

blue

D2

Generic S6

EMS

EMS

2

0.361

round

blue

D3

Suvvia

Sigma Pharma

EMS

4

0.363

round

white

D4

Sandoz

Sandoz

4

0.302

clover

blue

E

Generic S7

According to the leaflets accompanying the packages, all tablets from the same manufacturer, regardless of type (generic or brand-named) presented the same exact composition. For different manufacturers, small variations in tablets composition were observed, with one excipient/adjuvant added or replaced. Use of the automatic multi-component 7

ACCEPTED MANUSCRIPT

spectral matching (deconvolution) tool available from the SpectaT M software for the analysis of the averaged spectral profile of each type of tablet successfully detected sildenafil

PT

(spectral bulk) and 4 more components (excipients/adjuvants) for each manufacturer. All

RI

substances detected by this approach were consistent with leaflet information (Table 2). Although no details concerning the deconvolution algorithm employed by the SpectaT M

SC

software were found on the documentation available, analysis of the spectral profiles of each substance detected revealed specific, characteristic spectral bands, which were also

NU

present in the spectral profile of the tablets. The position of these bands, found both outside and inside the fingerprint region, was used during objective visual comparison of spectral

MA

profiles obtained from the tablets to identify and explain the small differences between samples from different manufacturers.

D

Table 2: Sildenafil-based tablets composition by manufacturer (according to the leaflets) and results from

TE

multi-component (deconvolution) analysis. Components were abbreviated as sld (sildenafil), mcc (microcrystalline cellulose), cpd (calcium phosphate dibasic), mgs (magnesium stearate), lac (lactose), croscarmel-

AC CE P

lose sodium (ccs), silicon dioxide (sdo) and crospovidone (cpv). Manufacturer

Composition (leaflets)

Components Detected

Group

Pfizer

sld, mcc, cpd, ccs, mgs

sld, mcc, ccs, cpd, mgs

A

Eurofarma

sld, mcc, sdo, ccs, cpd, mgs

sld, cpd, mcc, ccs, mgs

B

Brainfarma

sld, mcc, cpd, ccs, mgs

sld, mcc, ccs, cpd, mgs

C

EMS

sld, mcc, lac, ccs, sdo, mgs

sld, mcc, lac, ccs, mgs

D

Sandoz

sld, mcc, cpd, cpv, ccs, mgs

sld, mcc, ccs, cpd, mgs

E

As expected, due to the similar composition, the averaged spectral profiles for all types of sildenafil-based tablets studied presented a high degree of similarity, appearing to be indistinguishable from one another (Fig. 1). A closer inspection on the overlayed profiles, however, revealed that they were, in fact, separated into groups which corresponded to tablets produced by the same manufacturer, regardless of the brand name, the pharmaceu8

ACCEPTED MANUSCRIPT

tical company or if the tablets were brand-named or generic.

PT

Figure 1: Averaged, overlayed, spectral profiles for the 13 types of sildenafil-based tablets listed on Table 1.

AC CE P

TE

D

MA

NU

SC

RI

Strong contributions from sildenafil can be seen at 1698cm−1 , 1171cm−1 , 939cm−1 and 734cm−1 .

The separation was more evident at specific bands, each attributed to contributions from one excipient/adjuvant present in each mixture. Although, as shown in Table 2, the same components are present in most tablets, their proportions in the mixture are not required to nor need to be the same (except for the AI, sildenafil), allowing for variations in intensity for each band, from manufacturer to manufacturer. The spectral contributions of each component can be better perceived and understood by first examining the individual spectral profiles of the most significative components (Figs. 2 to 9). Closer inspection of the overlayed profiles, focused at region 1 (3600 − 2700cm−1 ) (Fig. 10), revealed the presence of peaks centered at 3553cm−1 and 3472cm−1 , attributed to calcium phosphate dibasic (Fig. 9), appearing very clearly in 3 of the 13 overlayed profiles, 9

ACCEPTED MANUSCRIPT

TE

D

MA

NU

SC

RI

PT

Figure 2: Sildenafil spectral profile.

AC CE P

Figure 3: Tadalafil spectral profile.

10

ACCEPTED MANUSCRIPT

TE

D

MA

NU

SC

RI

PT

Figure 4: Microcrystalline cellulose spectral profile.

AC CE P

Figure 5: Croscarmelose Sodium spectral profile.

11

ACCEPTED MANUSCRIPT

TE

D

MA

NU

SC

RI

PT

Figure 6: Crospovidone spectral profile.

AC CE P

Figure 7: Lactose spectral profile.

12

ACCEPTED MANUSCRIPT

TE

D

MA

NU

SC

RI

PT

Figure 8: Magnesium stearate spectral profile.

AC CE P

Figure 9: Calcium phosphate dibasic spectral profile.

13

ACCEPTED MANUSCRIPT

all of them corresponding to samples from group B (manufacturer Eurofarma). Traces of the same peaks were visible in 2 other profiles, both corresponding to samples from the

PT

group C (manufacturer Brainfarma). The same peaks could not be visually identified in

RI

the remaining profiles. The presence of these peaks separating groups B and C from the others indicates a larger concentration of calcium phosphate dibasic in samples from

SC

group B, followed by group C, while groups A, D and E appear to have lower (below visual detection) concentration or no presence of this component in their formulation.

NU

Still in the same region, other two peaks, centered at 2915cm−1 and 2849cm−1 , attributed to magnesium stearate (Fig. 8), were unmistakably detected in all profiles. Rel-

MA

ative intensities of these peaks were visibly higher for 2 profiles, both corresponding to group C, indicating a probable higher concentration for this compound in tablets produced

D

by this manufacturer (Brainfarma). Another very important peak was present at 2900cm−1 ,

TE

between the magnesium stearate peak pair. This peak, attributed to lactose (Fig. 7), was present in 4 profiles, all of which corresponded to tablets of group D (manufacturer EMS),

AC CE P

thus separating this group from all the others. At region 2 (1250 − 950cm−1 ) (Fig. 11), separation of Group A (manufacturer Pfizer) accompanied the valley centered around 1143cm−1 , attributed to microcrystalline cellulose (Fig. 4), while profiles from other groups presented stable or increasing intensity, indicating a probable higher proportion of this compound, relative to other excipients/adjuvants, compared to other manufacturers. Similarly, separation of group E (manufacturer Sandoz), occurring between 1075cm−1 and 975cm−1 , was attributed to the presence of crospovidone (Fig. 6), which presented lower signal intensity in this band. Separation of group B was also observed in this region, due to the higher intensity peaks at 1054cm−1 and 983cm−1 , once again attributed to the higher concentration of calcium phosphate dibasic for this manufacturer (Eurofarma). Separation between the same groups were detected at other regions of Fig. 1, varying 14

ACCEPTED MANUSCRIPT

TE

D

MA

NU

SC

RI

PT

Figure 10: Zoom at region 1 (3600 − 2700cm−1 ) of Fig. 1.

AC CE P

Figure 11: Zoom at region 2 (1250 − 950cm−1 ) of Fig. 1.

15

ACCEPTED MANUSCRIPT

Table 3: Correlations between averaged spectral profiles for all genuine sildenafil-based tablet brands listed on Table 1. Blue cells (higher values) show correlations between tablets within the same group (same

PT

manufacturer). Yellow cells show the lowest values obtained, corresponding to correlation between groups B (Eurofarma) and D (EMS). Cyan cells indicate the highest correlations obtained for samples from different

RI

groups (A and E). The highest correlation value obtained (99.6%), corresponding to correlation between

group B2 and D3, is colored in red. A2

A3

B1

B2

1

A2

0.990

1

A3

0.993

0.988

1

B1

0.921

0.921

0.920

1

B2

0.918

0.916

0.915

0.996

B3

0.925

0.923

0.925

0.991

C1

0.975

0.967

0.974

0.918

C2

0.980

0.973

0.977

D1

0.956

0.954

0.954

D2

0.947

0.946

0.946

D3

0.932

0.927

D4

0.956

E

0.985

C1

C2

D1

D2

D3

D4

E

NU

A1

B3

MA

A1

SC

group B1 and B2, is colored in green. The lowest value (84.7%), corresponding to correlation between

1

D

0.988

TE

0.913

1

0.921

1

0.918

0.927

0.989

1

0.877

0.870

0.882

0.948

0.951

1

0.873

0.864

0.877

0.937

0.945

0.989

1

0.936

0.858

0.847

0.864

0.927

0.928

0.984

0.983

1

0.955

0.956

0.877

0.868

0.881

0.953

0.953

0.994

0.989

0.986

1

0.983

0.985

0.903

0.898

0.913

0.981

0.984

0.952

0.948

0.931

0.957

AC CE P

0.922

from plainly visible to more subtle levels, but always related to the components and their relative proportions in the tablets from each manufacturer. In the final step of analysis, correlation between the averaged spectral profiles of the 13 types of genuine tablets studied was evaluated using the QCheck tool available from the OMNICT M software. Both the overall visual similarity and the separation into spectral groups according to manufacturer were reproduced by the spectral correlation analysis (Table 3). Highest correlation values (up to 99.6%) were obtained between samples within 16

1

ACCEPTED MANUSCRIPT

the same group (same manufacturer) and lowest (down to 84.7%) between samples from different groups (different manufacturers), thus reproducing exactly the same behavior

PT

observed previously.

RI

3.1.2. Tadalafil tablets

SC

The nominal and physical characteristics obtained for the genuine tadalafil-based samples studied were resumed on Table 4.

NU

Table 4: Nominal and physical characteristics of all genuine 20mg tadalafil-based tablets studied. The mass

Brand

Company

Manufacturer

Mass (g)

Shape

Color

Group

Cialis

Lilly

Eli Lilly

4

0.361

oval

yellow

F1

Ciavor

Biolab

Eli Lilly

2

0.369

oval

yellow

F2

Generic T1

Sandoz

Eli Lilly

2

0.364

oval

yellow

F3

Zyad

Ach´e

Biossint´etica

2

0.368

round

yellow

G1

Generic T2

Biossint´etica

Biossint´etica

2

0.366

round

yellow

G2

Tada

Eurofarma

Eurofarma

2

0.368

oval

white

H1

AC CE P

TE

MA

Samples

D

column contains the average mass per tablet of each brand.

Generic T3

Eurofarma

Eurofarma

2

0.366

oval

white

H2

Generic T4

Medley

Medley

5

0.367

round

yellow

I

Generic T5

Prati Donaduzzi

Prati Donaduzzi

2

0.363

round

yellow

J

According to the leaflets accompanying the packages, all samples from the same manufacturer, regardless of type (generic or brand-named) presented the same exact composition. For different manufacturers, the only apparent change was the addition of silicon dioxide for tablets manufactured by Eurofarma (group H). Multi-component/deconvolution analysis successfully detected lactose (spectral bulk), tadalafil and microcrystalline cellulose for all tadalafil-based groups, except H, for which only lactose and tadalafil were detected. Nevertheless, all substances detected by this approach were consistent with leaflet 17

ACCEPTED MANUSCRIPT

information (Table 5). Detection of a smaller number of components by deconvolution, despite tadalafil-based tablets leaflets listed the presence of as many excipients/adjuvants as

PT

on sildenafil-based tablets, was attributed to a much larger proportion of lactose, seconded

RI

by tadalafil and microcrystalline cellulose, when compared to other components, causing the tablets spectral profile to be dominated by the contributions of these components for

SC

most tablets tested.

Table 5: Tadalafil-based tablets composition by manufacturer (according to the leaflets) and results from

NU

multi-component (deconvolution) analysis. Components were abbreviated as tad (tadalafil), lac (lactose), hpc (hidroxypropylcellulose), mcc (microcrystalline cellulose), sls (sodium laurilsulfate), mgs (magnesium

MA

stearate), croscarmellose sodium (ccs) and silicon dioxide (sdo). Composition (leaflets)

Components Detected

Group

Eli Lilly

tad, lac, ccs, hpc, mcc, sls, mgs

tad, lac, mcc

F

Biossint´etica

tad, lac, hpc, ccs, sls, mcc, mgs

tad, lac, mcc

G

Eurofarma

tad, lac, hpc, sls, ccs, mcc, sdo, mgs

tad, lac

H

Medley

tad, mcc, lac, hpc, ccs, sls, mgs

tad, lac, mcc

I

Prati Donaduzzi

tad, lac, ccs, sls, mcc, mgs

tad, lac, mcc

J

AC CE P

TE

D

Manufacturer

As observed for sildenafil-based tablets, despite the differences in shape and color, the similar composition of all genuine tadalafil-based tablets was visually evidenced from the overlayed spectra (Fig. 12).

Differently from results obtained from sildenafil-based tablets, no significative group separation was verified between different tablet manufacturers, except for samples from group I, whose spectral profile could be clearly distinguished from all others at specific regions (Fig. 13). The separation was attributed to an increase in microcrystalline cellulose proportion relative to lactose when compared to other manufacturers, thus reducing signal intensity for the lactose band around 760cm−1 . The higher degree of similarity shared by all 9 types of genuine tadalafil-based tablets, 18

ACCEPTED MANUSCRIPT

Figure 12: Averaged, overlayed, spectral profiles for the 9 types of tadalafil-based tablets listed on Table 4. The lactose (Fig. 7) dominance is self-evident. Tadalafil (Fig. 3) contributions can be seen at 1672cm−1 ,

TE

D

MA

NU

SC

RI

PT

1639cm−1 and 744cm−1 .

Figure 13: Zoom of the overlayed averaged standard spectral profiles of tadalafil-based tablets studied.

AC CE P

Separation of group I (in red) was the only to be clearly distinguishable from all others when overlayed.

19

ACCEPTED MANUSCRIPT

attributed to lactose spectral dominance, was confirmed by the results from correlation analysis (Table 6). Correlations intra- and extra-group presented high values, from 98, 4%

RI

thus confirming separation of this group from the others.

PT

to 99, 6%, except for group I, which presented the lowest values, from 90, 2% to 93, 0%,

Table 6: Correlations between averaged spectral profiles for all genuine tadalafil-based tablet brands listed

SC

on Table 4. Blue cells show correlations between tablets within the same group (same manufacturer). Yellow cells show the lowest values obtained, corresponding to correlation between group I (Medley) and all the

NU

others. The highest value obtained (99.6%), corresponding to correlation between group F1 and J, is colored in green. The lowest value (90.2%), corresponding to correlation between group H2 and I, is colored in red. G1

MA

F3

G2

H1

F1

1

F2

0.984

1

F3

0.993

0.986

1

G1

0.994

0.989

D

F2

1

G2

0.995

0.987

0.992

0.992

1

H1

0.989

0.984

0.991

0.990

0.987

1

AC CE P

F1

TE

0.994

H2

I

H2

0.989

0.992

0.991

0.991

0.991

0.993

1

I

0.921

0.905

0.923

0.930

0.912

0.914

0.902

1

J

0.996

0.988

0.993

0.993

0.994

0.991

0.992

0.907

J

1

At this point, characterization results obtained from genuine samples confirm the similarity, at least from the infrared spectral point of view, between brand-named and generic tablets bearing the same AI, as expected. For tablets made by the same manufacturer, regardless of the pharmaceutical company or if the tablets were sold as generic or beared some specific brand name, all samples became indistinguishable once the external layer was removed and the core crushed. Despite the strong similarities, sildenafil-based samples from different manufacturers studied were nevertheless found to be distinguishable from one another, due to small variations on their spectral profiles, attributed to slight 20

ACCEPTED MANUSCRIPT

variations in excipients/adjuvants proportions. The same degree of variation, however, was not observed among tadalafil-based samples, for which all tablets, from all manufac-

PT

turers studied, could not be clearly distinguished from one another, except from samples

RI

from group I, whose formulation apparently contained a higher amount of microcrystalline cellulose. Such difference between sildenafil- and tadalafil-based tablets could be possibly

SC

explained if we consider that, as usual, each AI entered the market as a single, exclusive, patent-protected, medicine: Viagra (sildenafil) and Cialis (tadalafil). It was only

NU

after patent rights expired that other manufacturers were allowed to produce and market sildenafil- and tadalafil-based medicines, thus flooding drugstores with several different

MA

options for each type of PDE5 inhibitor, both generic and brand-named. Before entering the market, however, each new product is required to pass bioequivalence tests when

D

compared to a previously approved medicine (in this case, Viagra or Cialis). Such require-

TE

ment explains why all tablets based on the same AI presented such similar compositions, regardless of manufacturer. Considering that patent rights for Viagra expired in Brazil

AC CE P

in 2010, while Cialis expired only in 2015, each manufacturer now producing sildenafilbased tablets, putatively bioequivalent to Viagra, had considerably more time (5 years) to adjust/optimize its own formulation, while still keeping bioequivalency to Viagra, but creating room for the slight variations observed in the present study. As patent rights for Cialis just recently expired, such differentiation process is possibly still in a very early stage, thus justifying why most tadalafil-based tablets available on the Brazilian market are virtually spectrally indistinguishable from the reference medicine, Cialis. 3.2. Seized tablets Once characterization of the genuine tablets was completed, a number of questioned tablets, seized by local law enforcement forces on different occasions, was submitted to the same tests, allowing comparison between genuine and questioned merchandise. Ques21

ACCEPTED MANUSCRIPT

tioned tablets properly packaged or bearing identification marks were compared to the correspondent genuine tablet brand/group standard previously studied (Tables 1 and 4.

PT

For questioned tablets with no packaging nor identification marks, but still suspected to be

RI

PDE5 inhibitors, best matches for composition (deconvolution), objective visual spectral profile comparison and correlation among all standards available were used to provide a

SC

probable manufacturer identification.

NU

3.2.1. Sildenafil tablets

Questioned sildenafil tablets consisted of 12 tablets packaged as generic 50mg silde-

MA

nafil from Neo Qu´ımica pharmaceutical company (reference: group C2 , Table 1), divided into 8 groups (6 different batches/lot numbers, 2 unknown), 12 tablets packaged as generic 50mg sildenafil from Eurofarma pharmaceutical company (reference: group B2 ), divided

D

into 5 groups (4 different batches/lot numbers, 1 unknown), 1 tablet packaged as Dejav´u

TE

(reference: group B1 ), and 13 tablets packaged as Viagra (reference: group A1 ), divided into 6 groups (4 different batches/lot numbers, 2 unknowns).

AC CE P

Seized tablets bearing Neo Qu´ımica identification marks presented no divergence regarding physical traits, specially tablet color and mass. Spectral deconvolution identified the same composition of genuine tablets (Table 2), while objective visual spectral profile comparison, taken as an indication of tablet components proportions, matched group C (manufacturer: Brainfarma). The highest correlation values for these samples, which oscillated from 98, 1% to 99, 2%, also corresponding to group C. The convergence of all tests (data not shown) to the same group, matching the tablet manufacturer indicated by the packaging, allowed these samples to be reported as genuine. Seized tablets bearing Eurofarma identification marks also presented no divergence regarding physical traits, specially tablet color and mass. Spectral deconvolution identified the same basic main composition of genuine tablets, while objective visual spectral profile 22

ACCEPTED MANUSCRIPT

comparison matched group B (manufacturer: Eurofarma). The highest correlation values for these samples, which oscillated from 95, 9% to 98, 9%, also corresponded to group B.

PT

Once more, the convergente of all tests (data now shown) to the same group, matching the

RI

tablet manufacturer indicated by the packaging, allowed these samples to be reported as genuine.

SC

For seized tablets labeled as Viagra, however, several discrepancies were verified on all tests performed when compared to reference standard (group A1 , Table 1), starting with

NU

color and mass variations. Despite multi-component/deconvolution analysis was able to detect sildenafil in all questioned tablets, its spectral contribution to the profiles of many

MA

samples was weaker, evidencing smaller proportions of sildenafil for such tablets. Foreign components, not meant to be present on the tablets (not listed in leaflets), such as

D

starch and plaster of paris (calcium sulfate), were detected presenting strong contributions

TE

to the spectral profiles of many samples. Starch was detected in 8 samples and plaster of paris was found in 6. Also, a very strong contribution of lactose to the spectral

AC CE P

profile, when compared to genuine Viagra standards, was observed in 4 tablets. Multicomponent/deconvolution final results detected significative compositional divergence in 10 of the 13 questioned Viagra tablets tested. The compositional divergence among questioned Viagra tablets was also clearly detectable by objective visual inspection of the spectral profiles of the samples. Six different profile types (one for each group of questioned tablets) were observed (Fig. 14). Only samples from group K (Table 7) visually matched genuine Viagra. As could be expected from such variation on the spectral profiles, correlation values between questioned and genuine Viagra profiles reached values as low as 3.9% for some samples and as high as 98.6% for others. The final results for seized Viagra samples were resumed on Table 7. Of the 13 seized tablets packaged as Viagra, 10 presented enough divergence from genuine samples on all 23

SC

RI

PT

ACCEPTED MANUSCRIPT

Figure 14: Spectral profiles obtained from seized Viagra tablets. Six different profile types (one for each

NU

seized group) were observed. Although multi-component/deconvolution analysis detected sildenafil in all

AC CE P

TE

D

MA

profiles, only samples from group K (profile shown in red) matched genuine Viagra.

24

ACCEPTED MANUSCRIPT

Table 7: Final results for the analysis of seized Viagra samples. Compositional results, obtained from spectral deconvolution were resumed as sld (sildenafil), mcc (microcrystalline cellulose), ccs (croscarmelose

PT

sodium), cpd (calcium phosphate dibasic), mgs (magnesium stearate), lac (lactose), ptr (plaster of paris) and

Samples

Mass (g)

K

L20483005K

3

0.315

L

9619710B

2

M

314833021B

N

Composition

Result

sld, mcc, ccs, cpd, mgs

0.986

genuine

0.302

sld, mcc, lac, mgs

0.803

counterfeit

2

0.417

ptr, sld, stc

0.669

counterfeit

L13621\6\15

3

0.420

ptr, sld, stc

0.602

counterfeit

O

unidentified

1

0.413

ptr, sld, stc

0.483

counterfeit

P

L904830028

2

0.353

lac, stc, mcc, sld

0.039

counterfeit

SC

Correlation

NU

Lot/Batch #

MA

Group

RI

stc (starch).

D

tests performed to be reported as counterfeits. The remaining 3 tablets (group K), on the

be reported as genuine.

AC CE P

3.2.2. Tadalafil tablets

TE

other hand, presented high convergence to standard Viagra samples, thus allowing them to

A total of 12 tablets packaged as Cialis, divided into 5 groups (3 different batches/lot numbers, 2 unknowns) were available among seized samples. As observed for seized Viagra tablets, several discrepancies were verified on all tests performed when compared with reference standards (group F1 , Table 4), starting with color and mass variations. Multi-component/deconvolution detected tadalafil only in 6 tablets, 2 of which presenting a much weaker signal for tadalafil compared to the standard profile, suggesting lower dosage. Sildenafil was detected in 6 tablets. Lactose was found in 8 tablets. Microcrystalline cellulose, calcium phosphate dibasic and starch were also detected. Objective visual spectral comparison detected 4 different profile types, because 2 of the 5 seized groups presented the same behavior when overlayed, as shown in Fig. 15.

25

ACCEPTED MANUSCRIPT

Figure 15: Spectral profiles obtained from questioned Cialis tablets. Four different profile types where observed, because 2 of the 5 seized groups presented similar (closely superposed) profiles. Only samples

TE

D

MA

NU

SC

RI

PT

from group U (profile shown in red) matched genuine Cialis.

AC CE P

Only samples from group U (Table 5) visually matched genuine Cialis. For the correlation tests, a high level of dominance from the spectral contribution of lactose was observed on many questioned samples, like previously observed for genuine samples. This resulted in high correlation values even for samples in which tadalafil was not detected, reaching up to 97.3%. Confront with results from Table 6, however, showed that, despite the lactose dominance, correlation between all genuine samples studied in which lactose and tadalafil were present as major components reached minimum level at 98.4%, even between samples from different manufacturers (excluding those from Group I, where lactose, tadalafil and microcrystalline cellulose were identified as the main components). Therefore, the high content of lactose in genuine samples explained why questioned samples whose spectral profiles presented strong contribution from lactose (but none from tadalafil) could reach high correlation values to genuine samples but, nevertheless, lower than those from sam26

ACCEPTED MANUSCRIPT

ples where both lactose and tadalafil were detected on the spectral profile.

PT

Table 8: Final results for the analysis of seized Cialis samples. Compositional results, obtained from spectral deconvolution were resumed as tad (tadalafil), lac (lactose), sld (sildenafil), mcc (microcrystalline cellulose),

Lot/Batch #

Samples

Mass (g)

Composition

SC

Group

RI

cpd (calcium phosphate dibasic) and stc (starch).

Correlation

Result

LA10309

4

0.449

sld, mcc, cpd, stc

0.078

counterfeit

R

L806397

2

0.413

lac, mcc, sld, tad

0.734

counterfeit

S

unidentified

1

0.338

lac, stc

0.971

counterfeit

T

L802513

1

0.346

lac, mcc

0.973

counterfeit

U

unidentified

4

0.364

lac, tad

0.993

genuine

MA

NU

Q

Final results from seized Cialis samples were resumed on Table 8. Of the 12 seized

D

tablets packaged as Cialis, 8 presented enough divergence from genuine samples on all

TE

tests performed to be reported as counterfeits. The remaining 4 tablets (group U), on the other hand, presented high convergence to standard Cialis samples, thus allowing them to

AC CE P

be reported as genuine.

3.2.3. Unmarked/Unindentified tablets In last stage of the present work, 15 seized samples bearing totally unrecognizable or no identification marks, but still suspected to contain sildenafil or tadalafil, were submitted to the same analysis previously described. Most of these samples were received partially or totally crushed, preventing the external layer to be removed prior to the analysis. The spectral profiles of each of these samples were submitted to deconvolution, objetive visual comparison and correlation tests. The multi-component/deconvolution analysis was employed to identify the AI and main excipients/adjuvants present on each sample. Objetive Visual comparison, guided by the results from deconvolution, and correlation to all reference groups previously identified 27

ACCEPTED MANUSCRIPT

(Tables 1 and 4) were then performed. Best visual correspondence group was selected by overlaying the spectral profiles of each sample to the averaged standard profiles showed

PT

on Figures 1 and 12 and matching group characteristics like those exemplified on Fig-

RI

ures 10, 11 and 13. If no reasonable visual match was found, the sample was considered to match an unknown group. Best correlation group was selected from the highest correlation

SC

value between each sample and all available averaged standards. Correlation values from genuine samples within the same group (Tables 3 and 6) were used as a reference.

NU

The final result for each unmarked/unidentified sample was evaluated from the agreement between the results from all tests performed and is resumed in Table 9. Inconclu-

MA

sive final results concerning the manufacturer were obtained for 5 of the 15 unidentified samples (3 containing sildenafil and 2 containing tadalafil). It is possible that these sam-

D

ples represented examples of irregular/unlicensed (but not necessarily counterfeited) PDE5

TE

tablets, manufactured outside Brazil and sold illegally on the local market, like Pramil, Potenciem, Enermil, Digram and other brand names, which were not previously character-

AC CE P

ized in this study. Counterfeit final results were obtained for 2 of the unidentified samples, which presented individual test results comparable to those obtained for some samples identified as counterfeit Cialis. Analysis of all questioned seized samples available for the present study showed that counterfeit samples packaged as Viagra and Cialis can be easily detected by the proposed approaches, as sample characteristics studied, both physical and spectral, presented a high level of divergence from genuine samples. Although not present among available seized samples, counterfeits of PDE5 samples other than Viagra and Cialis, a possible reality in the near future, would be detected much in the same way.

28

PT

ACCEPTED MANUSCRIPT

RI

Table 9: Final results from the analysis of unmarked/unindentified samples. Compositional results, obtained

SC

from spectral deconvolution were resumed as tad (tadalafil), lac (lactose), sld (sildenafil), mcc (microcrystalline cellulose), cpd (calcium phosphate dibasic), mgs (magnesium stearate) and stc (starch).

1

mcc, sld, cpd, mgs

B

NU

Final Result (Manufacturer)

2

mcc, sld, lac, mgs

D

0.968 (D)

Sildenafil (EMS)

3

cpd, sld

B

0.941 (B)

Sildenafil (Eurofarma)

4

sld, cpd, mgs

unknown

0.947 (C)

Sildenafil (unknown)

5

sld, cpd, mcc

D

Composition

Correlation

0.922 (B)

Sildenafil (Eurofarma)

6

sld, mcc, cpd, mgs

C

0.977 (C)

Sildenafil (Brainfarma)

7

sld, lac, mgs, stc

unknown

0.951 (C)

Sildenafil (unknown)

8

sld, mcc, cpd, mgs

A

0.990 (A)

Sildenafil (Pfizer)

sld, lac, mgs, stc

unknown

0.949 (C)

Sildenafil (unknown)

sld, mcc, cpd, mgs

B

0.967 (B)

Sildenafil (Eurofarma)

sld, mcc, cpd, mgs

C

0.981 (C)

Sildenafil (Brainfarma)

tad, mcc, mgs

unknown

0.475 (I)

Tadalafil (unknown)

13

AC CE P

Visual Sample

lac, mcc, tad, mgs

unknown

0.918 (I)

Tadalafil (unknown)

14

lac, mcc, sld, tad

F-H,J

0.977 (F)

Tadalafil (Lilly counterfeit)

15

lac, sld, tad

F-H,J

0.979 (J)

Tadalafil (Lilly counterfeit)

9 10 11 12

(Group)

0.975 (B)

MA

Group

TE

B

29

Sildenafil (Eurofarma)

ACCEPTED MANUSCRIPT

4. Conclusions

PT

In this work, ATR-FTIR was used to characterize a variety of genuine, both generic and brand-named, tablets based on PDE5 inhibitors sildenafil and tadalafil, legally produced

RI

and marketed in Brazil. Although several studies based on infrared profiling[12, 14–20]

SC

have been published over this decade concerning identification of counterfeit samples of Viagra and Cialis, most of these works do not consider nor discuss the compositional

NU

characteristics of the samples, employing spectral profiling mostly as a data source for chemometrical analysis, an approach which, although very powerful and useful, is usually

MA

beyond practical use in most forensic laboratories. The present approach was based on more direct and comparative practical analyses of the spectral profiles of tablet cores, applied not only to identify the AI of each tablet, but also the major excipients/adjuvants

D

present. The results obtained demonstrated that the excipients/adjuvants contributions to

TE

the spectral profile of each sample played a key role not only in discriminating between genuine and counterfeit samples, but also made possible to discriminate between genuine

AC CE P

tablets from different manufacturers. Additionally, by expanding the study beyond Viagra and Cialis and encompassing many generic and similar brands of Viagra- and Cialis-like tablets, not only the strong similarity between all genuine, generic and brand-named, medicines based on the same AI was verified, but also the indistinguishability between generic and branded tablets from the same manufacturer as well, confirming the results obtained by Custers et al. [20], who observed that Viagra generics manufactured by Pfizer were misclassified as genuine Viagra by chemometrics as well. From the forensics point of view, aside from simply classifying a questioned tablet as genuine or counterfeited, the results presented demonstrate that careful comparative analysis of the spectral profile of samples could be potentially used not only to identify

30

ACCEPTED MANUSCRIPT

the AI and excipients/adjuvants present, but also indicate possible legal manufacturers of

seized at different locations and dates to specific sources.

RI

Acknowledgements

PT

totaly unknown/unindentified tablets or even trace back and tie up counterfeited samples

SC

The authors wish to thank the support from Fundo de Incentivo a` Pesquisa da Pontif´ıcia Universidade Cat´olica de Minas Gerais (FIP - PUC MINAS) and Instituto de Crim-

NU

inal´ıstica da Pol´ıcia Civil de Minas Gerais (IC - PCMG).

MA

References

[1] H. A. Ghofrani, I. H. Osterloh, F. Grimminger, Sildenafil: from angina to erectile

TE

(2006) 689–702.

D

dysfunction to pulmonary hypertension and beyond, Nat. Rev. Drug Discov. 5 (8)

AC CE P

[2] L. A. B. Leoni, G. S. Leite, R. B. Wichi, B. Rodrigues, Sildenafil: two decades of benefits or risks?, Aging Male 16 (3) (2013) 85–91. [3] A. Das, D. Durrant, F. N. Salloum, L. Xi, R. C. Kukreja, PDE5 inhibitors as therapeutics for heart disease, diabetes and cancer, Pharmacol. Therapeut. 147 (2015) 12–21.

[4] C. Sternitzke, Drug repurposing and the prior art patents of competitors, Drug Discov. Today 19 (12) (2014) 1841–1847. [5] I. Aytac¸, J. Mckinlay, R. Krane, The likely worldwide increase in erectile dysfunction between 1995 and 2025 and some possible policy consequences, BJU Int. 84 (1999) 50–56.

31

ACCEPTED MANUSCRIPT

[6] K. M. Smith, F. Romanelli, Recreational Use and Misuse of Phosphodiesterase 5

PT

Inhibitors, J. Am. Pharm. Assn. 45 (1) (2005) 63–75. [7] A. Simsek, V. Tugcu, P. Erturkuner, F. Alkan, E. Ozbek, A. I. Tasci, Effects of the

SC

Int. Urol. Nephrol. 46 (10) (2014) 1889–1893.

RI

recreational use of PDE5 inhibitors on the corpus cavernosum of young, healthy rats,

[8] T. Melnik, The psychological effects of recreational PDE5 inhibitor use, Nat. Rev.

NU

Urol. 9 (9) (2012) 479–480.

MA

[9] J. Ames, D. Z. Souza, Counterfeiting of drugs in Brazil, Rev. Sa´ude Publ. 46 (1) (2012) 154–159.

[10] R. L. Hurtado, M. C. Lasmar, Counterfeit and contraband drugs in Brazil: overview

TE

D

and prospects for preventing their use, Cad. Sa´ude P´ublica 30 (4) (2014) 891–895. [11] R. S. Ortiz, K. C. Mariotti, R. P. Limberger, P. Mayorga, Physical profile of counter-

AC CE P

feit tablets Viagra and Cialis, Braz. J. Pharm. Sci. 48 (3) (2012) 487–495. [12] D. Custers, S. Vandemoortele, J.-L. Bothy, J. O. D. Beer, P. Courselle, S. Apers, E. Deconinck, Physical profiling and IR spectroscopy: simple and effective methods to discriminate between genuine and counterfeit samples of Viagra and Cialis, Drug Test. Anal. 8 (3-4) (2016) 378–387. [13] R. S. Ortiz, K. C. Mariotti, M. H. Holzschuh, W. R. ao, R. P. Limberger, P. Mayorga, Profiling counterfeit Cialis, Viagra and analogs by UPLCMS, Forensic Sci. Int. 229 (2013) 13–20. [14] P.-Y. Sacr´e, E. Deconinck, T. D. Beer, P. Courselle, R. Vancauwenberghe, P. Chiap, J. Crommen, J. O. D. Beer, Comparison and combination of spectroscopic techniques 32

ACCEPTED MANUSCRIPT

for the detection of counterfeit medicines, J. Pharm. Biomed. Anal. 53 (2010) 445–

PT

453. [15] F. Been, Y. Roggo, K. Degardin, P. Esseiva, P. Margot, Profiling of counterfeit

RI

medicines by vibrational spectroscopy, Forensic Sci. Int. 211 (2011) 83–100.

SC

[16] R. S. Ortiz, K. C. Mariotti, B. Fank, R. P. Limberger, M. J. Anzanello, P. Mayorga, Counterfeit Cialis and Viagra fingerprinting by ATR-FTIR spectroscopy with

NU

chemometry: Can the same pharmaceutical powder mixture be used to falsify two

MA

medicines?, Forensic Sci. Int. 226 (2013) 282–289. [17] M. J. Anzanello, R. S. Ortiz, R. P. Limberger, P. Mayorga, A multivariate-based wavenumber selection method for classifying medicinesinto authentic or counterfeit

D

classes, J. Pharm. Biomed. Anal. 83 (2013) 209–214.

TE

[18] M. J. Anzanello, R. S. Ortiz, R. Limberger, K. Mariotti, A framework for selecting

AC CE P

analytical techniques in profiling authentic and counterfeit Viagra and Cialis, Forensic Sci. Int. 235 (2014) 1–7. [19] M. J. Anzanello, F. S. Fogliatto, R. S. Ortiz, R. Limberger, K. Mariotti, Selecting relevant Fourier transform infrared spectroscopy wavenumbers for clustering authentic and counterfeit drug samples, Sci. Justice 54 (2014) 363–368. [20] D. Custers, T. Cauwenbergh, J. Bothy, P. Courselle, J. D. Beer, S. Apers, E. Deconinck, ATR-FTIR spectroscopy and chemometrics: An interesting tool todiscriminate and characterize counterfeit medicines, J. Pharm. Biomed. Anal. 112 (2015) 181– 189.

33

ACCEPTED MANUSCRIPT

Highlights

PT

• Identification of active ingredient and major excipients present on each sample.

RI

• Samples from different manufacturers can be distinguished on most cases.

• Easy identification of counterfeit samples.

SC

• Samples from same manufacturer are indistinguishable.

AC CE P

TE

D

MA

NU

• Unknown/unidentified samples can be traced back to possible manufacturers.

34