Microbial Ethanol Production Experimental Study and Multivariate Evaluation

10
 Microbialethanolproduction:Experimentalstudyandmultivariateevaluation § Va ss il iki A. Boumba a, *,Vangeli s Economou b , Ni ko laos Kourkoumelis c , Panagi ot a Gous ia b , Chri ssan thy Papa dopo ulou b , The odore Vou gio uklakis a a Department of Forensic Medicine& Toxicology, Medical School, Universityof Ioannina, 45110Ioaninna, Greece b Department of Microbiology, MedicalSchool,Universityof Ioannina, 45110Ioaninna, Greece c Department of MedicalPhysics,Medical School,Universityof Ioannina,45110Ioaninna, Greece 1.Introduction Themicrobial formation of ethanol isawellrecognized complication intheproperinterpretation of postmortemethanol analysis results[1–3] . Ma nymicroorganisms potentially presentin adeadbodyarecapableof ethanol production. Atleast58species of bacteria, 17speciesof yeastsand24species of moldscan produceethanol aswellasothervolatiles throughvarious biosynthetic pathways [1,4,5] . However, onlyalimitednumberof microbial specieshavebeen reportedtoproduceethanol inexperimental studies[6–11] . Fr om theavailablerelevant literature themicrobial speciesstudiedsofar are:Candidatropicalisinblood[6], Candidaalbicansinblood[6,7] andinurine[8,9],Candidatropicanainblood[6]andinurine[8], Candida parapsilosisinblood[6]andinurine[9],Corynebacterium sp.inblood[6], Lactococcus garviaeinblood[10],Escherichiacoliin blood[6]andinurine[8,9],Candida glabratainurine[11], Enterococcus andKlebsiellaoxytocainurine[8], andKlebsiella  pneumoniaeandProteusmirabilisinurine[9]. Inparticular, thespeciesC.albicans,C.tropicalis , C.tropicana,C.  glabrata,C. paralsilosis, E.coli, K.oxytoca, Corynebacteriumsp., Enterococcus andL. garviaehavebeenidentied inpostmortem bloodandurine[6,8,10,11] ,whileonlyC.albicans, C. parapsilosis ,K.  pneumoniae,E.coli, Proteusmirabilis, andS.cerevisiaehavebeen experimentally inoculatedinhumanbloodorurine[7,9,12] . ForensicScienceInternational215(2012)189–198 AR TICL EINFO  Article history: Received30September2010 Re cei ved in re vi sed form 3 Ma rch 2011 Accepted7March2011 Avai labl e onl ine 5 Apri l 2011 Keywords: Post-mortemblood Ethanol Microbialethanolproduction Multivariatestatistics Fermentation Volatiles Higheralcohols Biomarker(s) ABSTR ACT Ethanol can be pr oduced from all the postmortem avail able substrates, thoug h wi th hi gher rat es and yi eld s from carbo hydr ate s, durin g the ear ly st ages of putre fac tio n. The so-call ed hi gher al cohols (1- pro pano l, isobut anol, 2-methy l-1-butanol and 3-methy l-2-butanol ) and 1-butanol could be produc ed, fr om al l the avai labl e po stmortem subs trat es. Howe ver, a quanti tati ve relati onship betwee n the produced et han ol and the potenti all y produced other alcohol s is stil l missi ng. The obj ecti ve of thisstudywas the development of a simple , mat hema tica l mod el whi ch could be abl e to app roxi mat e the microbial pro duc edethanolin cor relation withotherproducedalcohols.Theselected bact eri al speciesinclud ed two Gram+ spore- forming anae robic bac teri a andtwo (oneGram+ one Gram-) aerobic /fac ultativ e anaerobic bac teri a, all bei ng commoncommen salsof the digest ive trac t andcommon col oni zers of the cor pse.The sele cted bac teri al str ains , Esch erich ia coli ,Clostridium perfrigens,Clostridium sporogenes and Enter ococc us faecal is, were cult ur ed sepa rately at 25 C, for 30 da ys, under controlled anae robic conditi ons. The produc ed ethanoland the pre viouslyreferr ed alco hol s wer e det erminedin the cultur e medium in 24 h interval s. Us ing pa rtial least squares (P LS) regr essi on, the estimation of the relevance score for the avai labl e de scri pt or s es tabl ished the stat is ti ca l mo de l to assess the et hanol concentration produced by eac h st udied microbe. E. co li,C. per fri gens,and C. spo rogene s produced dif fer ent pat tern s of ethanol and other alc ohols, whil e E. faecali s pro duc ed negligibl e amountsof etha nol and hi gher alcohols. In constructi ng the mathematical mode ls to pr edict the pr oduc ed ethanol, 1- pro pano l, 1-butanol, and isobuta nol were signic ant for C. perfr igens and C. spor ogenes , whi le 1-butanol, 1-prop anol, and met hyl- but anol were signicant for E.coli .The app lic abil ity of thes e mod els wastested in mi crobial , ana ero bi c cult ures of normal human bl ood and pl asma at 25 C. The res ul ts indi cate that factors such as the type of mic robe species, the glucose contentand the mediumcompos iti on app arently af fec t the procedure of mi crobial et hanol , and ot her al cohols producti on. However, the models can be appl ied wit h accep tab le acc uracy and the y show potential for appl ication in real post mort em cas es. 2011Els evier Ir ela nd Ltd . Al l ri ghts reserved. § 48thAnnualMeetingof theInternational Associationof ForensicToxicologists (TIAFT). Jointmeetingwiththesocietyof ToxicologicalandForensicChemistry (GTFCh). * Cor res pon din gauthor.Tel.:+302651007724;fax:+302651007857. E-mailaddresses: [email protected], [email protected](V.A.Boumba). ContentslistsavailableatScienceDirect Forensic Science Int ernationa l journa l homepage: www.elsevier .com/locate/forsciint 0379-0738/$seefrontmatter2011ElsevierIrelandLtd.Allrightsreserved. doi:10.1016/j.forsciint.2011.03.003

description

Microbial Ethanol Production Experimental Study and Multivariate Evaluation

Transcript of Microbial Ethanol Production Experimental Study and Multivariate Evaluation

Page 1: Microbial Ethanol Production Experimental Study and Multivariate Evaluation

7/18/2019 Microbial Ethanol Production Experimental Study and Multivariate Evaluation

http://slidepdf.com/reader/full/microbial-ethanol-production-experimental-study-and-multivariate-evaluation 1/10

Microbial 

ethanol 

production: 

Experimental 

study 

and 

multivariate 

evaluation§

Vassiliki A. Boumba a,*,  Vangelis Economou b, Nikolaos Kourkoumelis c, Panagiota Gousiab,Chrissanthy Papadopouloub, Theodore Vougiouklakis a

aDepartment   of   Forensic   Medicine  &   Toxicology,  Medical  School,  University  of   Ioannina,  45110  Ioaninna,  GreecebDepartment   of   Microbiology,  Medical  School,  University  of   Ioannina,  45110  Ioaninna,  GreececDepartment   of   Medical  Physics,  Medical  School,  University  of   Ioannina,  45110  Ioaninna,  Greece

1.  Introduction

The  microbial  formation  of   ethanol   is  a  well  recognized

complication   in  the  proper   interpretation  of   postmortem   ethanol

analysis  results  [1–3]. Many microorganisms   potentially present  ina  dead  body  are  capable  of   ethanol   production.   At  least  58  species

of   bacteria,  17  species  of   yeasts  and  24  species  of   molds   can

produce   ethanol   as  well  as  other   volatiles  through  various

biosynthetic  pathways  [1,4,5].

However,  only   a  limited  number   of   microbial  species  have   been

reported   to  produce   ethanol   in  experimental  studies  [6–11]. From

the available  relevant  literature  the microbial  species  studied  so  far

are: 

Candida 

tropicalis  in  blood  [6],  

Candida 

albicans  in  blood  [6,7]

and  in  urine   [8,9],   Candida  tropicana  in  blood  [6]  and  in  urine   [8],Candida   parapsilosis  in  blood  [6]  and  in  urine   [9],  Corynebacterium

sp.  in  blood  [6], Lactococcus 

 garviae  in  blood  [10], 

Escherichia 

coli  in

blood  [6]  and  in  urine   [8,9], 

Candida 

 glabrata  in  urine   [11],

Enterococcus  and  Klebsiella  oxytoca  in  urine   [8], and  Klebsiella

 pneumoniae  and  Proteus  mirabilis  in  urine  [9].

In  particular,  the  species 

C. 

albicans,  

C. 

tropicalis, C. 

tropicana, 

C.

 glabrata, 

C. 

 paralsilosis, E. 

coli, K. 

oxytoca, Corynebacterium  sp.,

Enterococcus  and  L.    garviae  have   been  identified  in  postmortem

blood  and  urine  [6,8,10,11],   while  only  C.  albicans, C.  parapsilosis,   K.

 pneumoniae, 

E. 

coli, Proteus 

mirabilis, and 

S. 

cerevisiae  have   been

experimentally  inoculated   in  human   blood  or  urine   [7,9,12].

Forensic  Science  International  215  (2012)  189–198

A  R   T  I   C  L   E  I  N  F   O

 Article history:

Received  30  September  2010

Received in revised form 3 March 2011

Accepted  7  March  2011

Available online 5 April 2011

Keywords:

Post-mortem  blood

Ethanol

Microbial  ethanol  production

Multivariate  statistics

Fermentation

Volatiles

Higher  alcohols

Biomarker(s)

A   B  S  T  R   A   C  T

Ethanol can be produced from all the postmortem available substrates, though with higher rates andyields from carbohydrates, during the early stages of putrefaction. The so-called higher alcohols (1-

propanol, isobutanol, 2-methyl-1-butanol and 3-methyl-2-butanol) and 1-butanol could be produced,

from all the available postmortem substrates. However, a quantitative relationship between the

produced ethanol and the potentially produced other alcohols is still missing.

Theobjective of thisstudywas thedevelopmentof a simple, mathematicalmodelwhich could be able

to approximate themicrobialproduced ethanolin correlationwithotherproducedalcohols.The selected

bacterial speciesincluded twoGram+ spore-forminganaerobic bacteria andtwo (oneGram+oneGram-)

aerobic/facultative anaerobic bacteria, all being commoncommensalsof thedigestive tract andcommon

colonizersof the corpse.The selectedbacterial strains, Escherichia coli, 

Clostridium perfrigens, 

Clostridium

sporogenes and Enterococcus faecalis, were cultured separately at 25 8C, for 30days, under controlled

anaerobic conditions. Theproduced ethanoland thepreviouslyreferred alcohols were determinedin the

culture medium in 24h intervals. Using partial least squares (PLS) regression, the estimation of the

relevance score for the available descriptors established the statistical model to assess the ethanol

concentration produced by each studied microbe. E. coli, 

C. perfrigens, 

and C. sporogenes produced

differentpatterns of ethanolandother alcohols,while E. faecalis produced negligible amountsof ethanol

and higher alcohols. In constructing the mathematical models to predict the produced ethanol, 1-propanol, 1-butanol, and isobutanolwere significant for C. perfrigens and C. sporogenes, while 1-butanol,

1-propanol, andmethyl-butanol were significant forE. coli. 

The applicability of these modelswas tested

in microbial, anaerobic cultures of normal human blood and plasma at 25 8C. The results indicate that

factors such as thetype ofmicrobe species, theglucose contentand themediumcomposition apparently

affect the procedure of microbial ethanol, and other alcohols production. However, the models can be

applied with acceptable accuracy and they show potential for application in real postmortem cases.

2011 

Elsevier Ireland Ltd. All rights reserved.

§ 48th  Annual  Meeting  of   the  International  Association  of   Forensic  Toxicologists

(TIAFT).   Joint  meeting  with  the  society  of   Toxicological   and  Forensic  Chemistry

(GTFCh).

* Corresponding 

author. 

Tel.: 

+30 

26510 

07724; 

fax: 

+30 

26510 

07857.

E-mail  addresses:  [email protected], [email protected]  (V.A.  Boumba).

Contents 

lists 

available 

at 

ScienceDirect

Forensic Science International

jour nal h o mepage: w ww.elsev ier   .co m/loc  ate / fo  r sc  i int

0379-0738/$  –  see  front  matter 

  2011  Elsevier   Ireland  Ltd.  All  rights  reserved.

doi:10.1016/j.forsciint.2011.03.003

Page 2: Microbial Ethanol Production Experimental Study and Multivariate Evaluation

7/18/2019 Microbial Ethanol Production Experimental Study and Multivariate Evaluation

http://slidepdf.com/reader/full/microbial-ethanol-production-experimental-study-and-multivariate-evaluation 2/10

Page 3: Microbial Ethanol Production Experimental Study and Multivariate Evaluation

7/18/2019 Microbial Ethanol Production Experimental Study and Multivariate Evaluation

http://slidepdf.com/reader/full/microbial-ethanol-production-experimental-study-and-multivariate-evaluation 3/10

penetration  depth  of   20  mm. After  injection  syringe  was  flushed  with  helium  for

1.5 

min. 

Acquisition 

time 

was 

20 

min. 

standard 

GC–FID 

chromatogram 

is

provided  as   complimentary  data  (Fig.  1C).

In  10  mL   HS  vials   containing  0.50  g   ammonium  sulfate were  added  500  mL   of   the

calibration  solution  or   of   the  culture  medium  or  of   blood  culture  and  500  mL   of   the

internal  standard  solution.  The  vials   were  then  sealed  with  metal  crimp  caps   fitted

with  silicon   septa  and  were  put  to  the  HS  auto  sampler  for  analysis.  Ammonium

sulfate  was   added  to  increase  the  ionic  strength  of   the  solution.  Calibration  curves

were  constructed  for  each   analyte  in  six  concentration  levels   within  the  relevant

working  concentration  range  (Table  1).

 2.3.  Multivariate  statistical  analysis

Regression  analysis  was  employed  to  model  the  correlation  between  the

microbial  produced  ethanol  and  the  other  higher  alcohols.  The  experimental  data

were 

analyzed 

with 

multiple 

linear 

regression 

(MLR) 

methods 

to 

model 

the

relationship  between  ethanol  concentrations  as  the  dependent  variable  being  in

linear  correlation  with   the  concentrations  of   the  other  alcohols   (independent

variables).  Using  the  model  built,  we  calculated  each   descriptor’s  relevance  to  the

current  model  using  partial  least  squares  (PLS)  regression  based  on   the  PLS1

algorithm.  Since  MLR is  very  sensitive  to  outliers,  the  dimensionality  of   some

models  has  been  decreased  by  leaving  out  certain  independent  variables.  This  has

also  permitted  us  to  find  the  simplest  acceptable  solution  in  each   case.

3. 

Results 

and 

discussion

 3.1.  Microbial  anaerobic   cultures  in  BHI   culture  medium

C. 

 perfrigens, C. 

sporogenes, 

E. 

coli  and 

E. 

 faecalis  (former  known

as 

Streptococcus 

 faecalis)  have  been   reported   as  being  the  main

colonizers   in  corpses  and  in  parallel  the  main  ethanol   producers

[4].  C.   perfrigens  and  C.  sporogenes  are  Gram-positive,  rod-shaped,obligate  anaerobic,  spore-forming  bacteria  of   the  genus   Clostridia

widely  distributed  in  nature   and  also  in  the  intestinal  tract  of 

humans  and  other   vertebrates,  insects,  and  soil.  E.  coli  is  a  Gram

negative,  facultative  anaerobic  and  non-sporulating,   rod-shaped

bacterium  that  is  commonly   found  in  the  lower  intestine  of   warm-

blooded  organisms  (endotherms).  E. 

 faecalis  is  a  Gram-positive

commensal   bacterium,   facultative  anaerobic,  inhabiting  the

gastrointestinal  tracts   of   humans   and  other   mammals  [20].

In  Fig.  1  the  growth  curves   for  each   bacterial  strain  during   the

30  days   of   incubation   at  25 

8C  are  presented.   By  the  fifth   day  of 

incubation   all  strains  reached   the  stationary  phase  of   growth.  At

8C  no  growth  was  observed  for  all  studied  bacterial  strains,  as  it

was  expected  (not   shown)   [9,21].

 3.2. 

Method 

validation

In  this   study  a  HS-GC–FID  method  has  been  optimized,   in  order

to  determine   simultaneously  the  concentrations   of   ethanol,   1-

propanol,   1-butanol,   isobutanol,  2-methyl-1-butanol   and  3-

methyl-2-butanol,   in  the  culture   medium   and  in  human   blood

samples.  These  volatiles  were  baseline  separated  from   other

common   determinants   in  volatile  blood  analysis  (methanol,

acetaldehyde,  2-propanol,  acetone,   ethyl   acetate,  butanone,   2-

butanol).  Baseline  separation  of   2-methyl-1-butanol   (active-amylalcohol)   and  3-methyl-2-butanol   (isoamyl  alcohol)   is   not  feasible

on  polar  columns   typically  used  for  separation  of   alcohols   [22].

Therefore,   herein   they  are  determined   as  mixture  and  referred  as

methyl-butanol.   The  method  was  evaluated  for  selectivity,  limit  of 

detection  (LOD),   limit  of   quantitation   (LOQ),   linearity,  precision

and  accuracy.   Selectivity  was  accomplished   by  analyzing  six

different  blank  samples  and  the  matrix  effect  was  evaluated.  The

LODs  and  LOQs  for  each   analyte  were  determined   as  the  lowest

concentration   of   each   analyte  yielding  a  signal  to  noise  ratio  of   at

least 3:1  and 10:1  respectively.  LODs  and  LOQs  are  listed  in  Table 1.

Linearity  was  expressed  by  the  correlation  coefficient   (R2)  of   the

regression  line  calculated  by  the  method   of   least-squares  with  a

weighting  factor   of   1/ x2.  Correlation  coefficient   exceeded  0.999  for

each   analyte  (Table  1).  Each   calibrator  was  back  calculated  againstthe  total  curve.   Precision  and  accuracy   of   the  method   were

evaluated  by  analyzing  three   QC  samples,  one  at  the  lower

concentration,   one  at  the  higher  and  one  within  the  working  range

of   concentration   of   each  analyte.  Precision  was  expressed  as  the

relative  standard   deviation  (%  RSD)   and  was  lower  than   6.0%  for  all

analytes.  Accuracy   of   the  method   was  calculated  as  the  percent

difference  from  the  expected  concentration   (%  Er)  and  ranged  from

1.0%  to  2.5%.

The  applied  methodology  was  based  on  previously  reported

methods   [17–19]  and  met  acceptable   analytical  criteria  for  the

quantitative  determination  of   alcohols.  Acetonitrile  was  used  as

internal   standard  since  it  has  similar  physicochemical   properties

with  the  determined   volatiles;  it  is  neither  an  ingredient  of 

alcoholic 

beverages 

nor 

putrefactive 

product 

[5]; and 

it 

was 

alsoused   in  ethanol   analysis  in  the  past  [17]. The  LODs   and  LOQs  of   our

method   lie  between  the  values  of   previous   reported  methods  being

either   lower  [19]  or  higher  [18,23].

 3.3. 

 Alcohols 

determination

Under   the  applied  experimental   conditions   the  clostridia

species  and  E.  coli  produced   ethanol,   1-butanol   and  higher  alcohols

in  variable  quantities  (Tables  2–4).  The  higher   amounts   of   ethanol

were  produced   from 

C. 

sporogenes, followed  by 

E. 

coli  and 

C.

 perfrigens.  The  most   abundant   produced   other  alcohol   was  1-

butanol   for  C.  sporogenes  and  C.   perfringens, followed  from  1-

propanol   and  isobutanol,  while  for 

E. 

coli  was  1-propanol.   The  less

abundant 

produced 

volatile 

were 

methyl-1-butanol 

for 

all 

the

 Table  1

Validation  of   the  HS-GC–FID  volatiles   determination  method.

Volatile  LOD  (mg/dL)  LOQ   (mg/dL)  Working  range  (mg/dL)  R2

Ethanol  0.01  0.03  0.03–400  0.999

1-Propanol  0.02  0.04  0.04–32.0  0.999

Isobutanol  0.005  0.015  0.02–0.08  0.989

1-Butanol 

0.005 

0.01 

0.01–0.08 

0.995

3-Methyl-2-butanol  0.01  0.025  0.04–0.32  0.997

3-Methyl-1-butanol  0.01  0.025  0.04–0.32  0.999

1,E+02

1,E+04

1,E+06

1,E+08

1,E+10

1,E+12

1,E+14

302520151050

Days

   B  a  c   t  e  r   i  a   l  c  o  u  n   t  s   (   l  o  g   C   F   U   /  m   l   )

Clostridium sporogenes Clostridium perfringensEnterococcus faecalis Escherichia coli  

Fig.  1.  Growth  curves  for   each  microbe under  study  showing  the number  of   bacteria

during 

the 

incubation 

period 

of  

30 

days 

at 

25 

8

under 

anaerobic 

conditions.

V.A.   Boumba   et   al.  /   Forensic   Science  International   215  (2012)  189–198  191

Page 4: Microbial Ethanol Production Experimental Study and Multivariate Evaluation

7/18/2019 Microbial Ethanol Production Experimental Study and Multivariate Evaluation

http://slidepdf.com/reader/full/microbial-ethanol-production-experimental-study-and-multivariate-evaluation 4/10

studied  microbes.  E. 

 faecalis  produced   negligible  amounts   of 

ethanol   and  other  alcohols  indicating  that   under  the  applied

conditions   the  branched   fermentation  pathways  producing

alcohols   were   unfavorable  for  this   microbe.   Therefore,   the

statistical  evaluation  of   the  relevant  results  was  practical  only

for  the  species 

C. 

 perfigens, 

C. 

sporogenes  and 

E. 

coli  and  impractical

for  E.    faecalis.

The  higher  alcohols   are  metabolic  products   of   the  fermentation

mainly   of   amino  acids  which  are  used  by  microbes   as  a  nitrogen

source.   The  higher  alcohols,  under  the  applied  experimental

 Table  2

Volatiles  concentration  during  the  fermentation  period  of   C.   perfrigens  in   BHI  culture  medium  at  25  8C.

Days   Ethanol  (g/L)  1-Propanol  (mg/dL)  1-Butanol  (mg/dL)  Isobutanol  (mg/dL)  Methyl-butanol  (mg/dL)

0 0.01  0  0  0  0

1   0.01  0.02  0.04  0  0

2   0.02  0.04  0.08  0.01  0.02

0.06 

0.08 

0.16 

0.01 

0

4   0.05  0.15  0.44  0.03  0.03

5   0.08  0.2  0.57  0.04  0.03

0.07 

0.26 

0.78 

0.05 

0.057   0.11  0.31  0.72  0.07  0.04

8   0.13  0.32  0.99  0.08  0.06

9   0.09  0.39  0.72  0.07  0.04

10  0.12  0.40  0.86  0.07  0.06

11  0.13  0.46  1.08  0.10  0.07

12  0.14  0.50  1.35  0.11  0.07

13 

0.12 

0.51 

1.02 

0.10 

0.09

14  0.16  0.52  1.51  0.11  0.09

15  0.14  0.48  1.18  0.11  0.07

16 

0.13 

0.42 

1.21 

0.11 

0.05

17  0.14  0.55  1.26  0.11  0.04

18  0.13  0.51  1.23  0.10  0.07

19  0.14  0.52  1.27  0.09  0.07

20  0.14  0.53  1.31  0.10  0.06

21  0.14  0.51  1.25  0.11  0.06

22  0.15  0.55  1.30  0.11  0.07

23  0.14  0.63  1.20  0.10  0.08

24  0.15  0.65  1.16  0.13  0.07

25  0.14  0.55  0.98  0.13  0.07

26 

0.15 

0.64 

1.52 

0.14 

0.08

27  0.15  0.60  1.47  0.13  0.09

28  0.15  0.68  1.32  0.12  0.08

29  0.16  0.69  1.90  0.12  0.10

30  0.15  0.70  1.33  0.15  0.08

 Table  3

Volatiles  concentration  during  the  fermentation  period  of   C.  sporogenes  in  BHI  culture  medium  at  25 

8C.

Days   Ethanol  (g/L)  1-Propanol  (mg/dL)  1-Butanol  (mg/dL)  Isobutanol  (mg/dL)  Methyl-butanol  (mg/dL)

0 0.00 

0.04 

0.00 

0

1   0.10  2.42  0.43  0.70  0.07

2   0.73  2.67  0.70  0.72  0.10

3   0.76  4.59  2.07  1.53  0.18

4   0.85  5.07  2.31  1.76  0.22

5   0.87  5.73  3.56  2.58  0.24

6   0.85  5.67  3.25  2.30  0.23

7   0.84  5.82  3.72  2.56  0.25

8   0.78  6.20  3.81  2.40  0.21

9   0.71  5.06  3.22  2.24  0.20

10 

0.79 

6.55 

4.77 

3.03 

0.34

11  0.81  6.59  4.81  3.12  0.32

12  0.84  6.68  5.19  3.18  0.32

13 

0.75 

6.67 

6.35 

3.36 

0.39

14  0.77  7.10  6.45  3.51  0.38

15  0.81  8.16  7.59  4.36  0.44

16  0.78  7.50  6.98  4.57  0.36

17  0.81  8.12  8.85  5.07  0.57

18  0.79  7.81  8.25  4.33  0.52

19  0.76  7.62  8.74  3.99  0.53

20 

0.77 

7.67 

8.00 

4.10 

0.52

21  0.78  8.55  9.20  4.43  0.51

22  0.76  8.21  8.87  4.20  0.55

23 

0.66 

8.20 

9.35 

4.35 

0.52

24  0.70  8.47  9.70  4.67  0.55

25  0.70  8.64  10.2  5.38  0.55

26  0.73  8.69  10.5  4.82  0.55

27  0.76  8.88  9.57  4.75  0.56

28  0.73  9.36  11.1  4.90  0.55

29  0.67  10.5  10.2  5.50  0.55

30  0.72  10.2  11.9  6.10  0.57

V.A.  Boumba   et   al.  /   Forensic   Science  International   215  (2012)  189–198192

Page 5: Microbial Ethanol Production Experimental Study and Multivariate Evaluation

7/18/2019 Microbial Ethanol Production Experimental Study and Multivariate Evaluation

http://slidepdf.com/reader/full/microbial-ethanol-production-experimental-study-and-multivariate-evaluation 5/10

conditions,   were   generated  in  parallel  with  ethanol.   Furthermore,

the  relatively  high  amounts   of   1-butanol   produced   from   the

clostridia  species  (Tables  2  and  3) indicate  that  clostridia  use  the

fermentation  pathway  for  1-butanol   and  acetone   formation  to

ferment  carbohydrates  to  ethanol   and  to  a  lesser  extend  to  1-

butanol   as  previously  has  been  suggested  [5].The  amounts   of   ethanol   produced   by  each  microbe  were  quite

different  although  the  initial  glucose  content   of   the  medium   was

the  same  (200  mg/dL).  The  observed  differences  in  microbes’

ability  to  produce   ethanol   and  higher   alcohols  during   fermentation

probably  reflect   differences  in  their  metabolic  efficiency  under

anaerobic  conditions   in  respect  to  the  substrate  catabolism.

These  results  indicate  that  the  glucose  content   is  not  the  only

determinant   of   the  produced   ethanol   –  although  it  should   have  a

significant  role  in  the  procedure,   and  the  process  is  potentially

affected  by  other  factors  as  well.  The  most  obvious  factor   is  themicrobe   type   even  for  species becoming  of   the  same  genera  (herein

C.   pergrigens  and  sporogenes).  The  patterns   of   the  higher  alcohols

have   also  shown  significant   differences  between  the  different

microbes’  cultures   indicating  the  flexibility  offered  by  branched

 Table  4

Volatiles  concentration  during  the  fermentation  period  of   E.  coli  in   BHI  culture  medium  at  25  8C.

Days  Ethanol  (g/L)   1-Propanol  (mg/dL)  Isobutanol  (mg/dL)  1-Butanol  (mg/dL)  Methyl-butanol  (mg/dL)

0  0  0  0  0  0

0.5  0.18  0.06  0.01  0.01  0.07

1   0.36  0.09  0.02  0.04  0.06

0.42 

0.11 

0.02 

0.07 

0.07

3   0.43  0.18  0.02  0.08  0.09

4   0.44  0.24  0.02  0.09  0.09

0.46 

0.34 

0.03 

0.09 

0.086   0.45  0.37  0.03  0.09  0.09

7   0.44  0.38  0.03  0.10  0.11

8   0.46  0.43  0.03  0.11  0.09

9   0.48  0.46  0.02  0.09  0.09

10  0.47  0.45  0.03  0.10  0.11

11  0.47  0.51  0.02  0.10  0.09

12 

0.46 

0.54 

0.03 

0.12 

0.11

13  0.50  0.62  0.03  0.09  0.08

14  0.45  0.68  0.02  0.11  0.12

15 

0.49 

0.66 

0.02 

0.10 

0.08

16  0.49  0.72  0.02  0.12  0.12

17  0.52  0.66  0.03  0.10  0.11

18  0.46  0.69  0.02  0.10  0.10

19  0.46  0.76  0.03  0.10  0.10

20  0.50  0.70  0.04  0.10  0.08

21  0.51  0.79  0.03  0.13  0.08

22  0.56  0.81  0.02  0.16  0.08

23  0.46  1.02  0.08  0.16  0.09

24  0.53  0.92  0.07  0.13  0.10

25 

0.49 

0.92 

0.07 

0.11 

0.09

26  0.53  0.95  0.05  0.11  0.09

27  0.50  0.91  0.02  0.10  0.09

28  0.52  0.86  0.07  0.10  0.09

29  0.49  0.86  0.09  0.10  0.09

30  0.50  0.88  0.10  0.08  0.09

Fig.  2.  (A)  Four  descriptors  (4D)  model  and;  (B)  one  descriptor  (1D)  model  for  ethanol  production  by  C.   perfrigens.

V.A.   Boumba   et   al.  /   Forensic   Science  International   215  (2012)  189–198  193

Page 6: Microbial Ethanol Production Experimental Study and Multivariate Evaluation

7/18/2019 Microbial Ethanol Production Experimental Study and Multivariate Evaluation

http://slidepdf.com/reader/full/microbial-ethanol-production-experimental-study-and-multivariate-evaluation 6/10

fermentation  pathways  to  microbes,  so  as,  to  cover  their  needs   for

energy   and  redox   balance  [5].

 3.4.  Models  of   microbial  ethanol   production

 3.4.1. 

Models 

 for  

C. 

 perfrigens

The  initial  model   for 

C. 

 perfrigens  was  built  using  four

independent   variables,  namely   the  1-propanol,  1-butanol,   iso-

butanol   and  methyl-butanol   concentrations   respectively.  Then   the

less  significant  variable  (methyl-butanol   with  descriptor  power

1.50)  was  skipped  and  a  model   with  three   independent   variables

(1-propanol,   1-butanol,   isobutanol)  was  built  and,   so  on,  iso-

butanol  was  skipped  (with descriptor power 82.94)  resulting  to  the

creation   of   the model with  two  independent   variables  (1-propanol,

1-butanol   with  descriptor  power  100  and  87.17  respectively).  The

significance  of  1-propanol   as a descriptor   (descriptor power 100)  in

this   case  is  so  powerful   that   a  considerable  satisfactory  model

could  be  created  by  using   it  as  the  only   independent   variable.

Thereinafter  are  given  for  comparison  the  models  by  using  four

independent   variables  (Spearman  Rank  Correlation,   r =  0.95)

(Eq.   (1)) and  the  most   powered  descriptor   variable  (Spearman

Rank  Correlation,  r =  0.95)  (Eq.   (2)).  In  Fig.  2  the  graphs   and  the

statistical  parameters  of   the  relevant  models   described  by  the

Eqs.  (1)  and  (2)  are  showed:

Ethanol   ¼  0:08    1Propanol   þ  0:03    1Butanol   þ

0:30    Isobutanol     0:01    Methyl-butanol   þ  0:03

(1)

Ethanol   ¼  0:11  þ  0:047    1Propanol   (2)

 3.4.2.  Models   for   C.  sporogenes

The  initial  model   for  C.   sporogenes  was  built  using  four

independent   variables,  1-propanol,   1-butanol,   isobutanol  and

methyl-butanol   concentrations,   respectively.  Then  the  less  signifi-

cant  variable  (methyl-butanol   with  descriptor   power  21.95)  was

skipped  and  a  model   with  three   independent   variables  (1-

propanol,   1-butanol,   isobutanol)  was  built  and,   so  on,  isobutanol

was  skipped  (descriptor  power   30.32)  resulting  to  the  develop-

ment  of   the  model  with  two  independent   variables  (1-propanol,   1-

butanol   with  descriptor  power  100  and  56.01  respectively).

Thereinafter  are  given  for  comparison   the  models   by  using  four

independent   variables  (Spearman  Rank  Correlation,  

r =  0.55

(Eq.   (3)) and  the  two  most  powered  descriptor   variable  (Spearman

Rank   Correlation,  r =  0.39)  (Eq.   (4)). In  Fig.  3  the  graphs  and  the

statistical  parameters  of   the  relevant  models   described  by  the

Eqs.  (3)  and  (4)  are  showed:

Ethanol   ¼  0:16    1Propanol     0:07    1Butanol   þ

0:61    Methyl-butanol     0:07    Isobutanol   þ  0:05

(3)

Ethanol   ¼  0:15    1Propanol     0:14 

Isobutanol   þ  0:09

(4)

 3.4.3. 

Models 

 for  

E. 

coli

The  initial  model   for  E.  coli  was  built  using  four   independent

variables,  1-propanol,   1-butanol,   isobutanol   and  methyl-butanol

concentrations   respectively.  Then  the  less  significant  variable

(isobutanol  with  descriptor  power  7.65)  was  skipped  and  the

model   with  three   independent   variables  (1-propanol,   1-butanol,

methyl-butanol)   was  built  and,   so  on,  1-propanol   with  descriptor

power  28.14  was  skipped  resulting  to  the  development  of   the

model  with  two  independent   variables  (1-butanol, methyl-butanol

with  descriptor   power   100  and  53.68  respectively).

Thereinafter  are  given  for  comparison   the  models   by  using  four

independent   variables  (Spearman  Rank   Correlation,  

r =  0.70)

(Eq.   (5)) and  the  two  most  powered  descriptors  (Spearman  Rank

Correlation,   r =  0.56)  (Eq.   (6)). In  Fig.  4  the  graphs   and  the

statistical  parameters  of   the  relevant  models   described  by  the

Eqs.  (5)  and  (6)  are  showed:

Ethanol   ¼  0:07    1Propanol   þ  0:20    Isobutanol   þ

1:61    1Butanol   þ  1:15    Methyl-butanol   þ  0:15

(5)

Ethanol   ¼  2:25    1Butanol   þ  0:98 

Methyl-butanol   þ  0:15

(6)

 3.4.4.   Aspects  of   ethanol   production  modeling 

Interestingly  the  significance  of   each   higher   alcohol   as  a

descriptor   in building  the  relevant model  was not  in  relevance with

its  produced   amount.   The  model  corresponding   to  C.   perfrigens  for

predicting   ethanol  production   (Eq.   (1))was highly  correlated  to  the

Fig. 

3. 

(A) 

Four 

descriptors 

model 

and; 

(B) 

two 

descriptor 

model 

for 

ethanol 

production 

by 

C. 

sporogenes.

V.A.  Boumba   et   al.  /   Forensic   Science  International   215  (2012)  189–198194

Page 7: Microbial Ethanol Production Experimental Study and Multivariate Evaluation

7/18/2019 Microbial Ethanol Production Experimental Study and Multivariate Evaluation

http://slidepdf.com/reader/full/microbial-ethanol-production-experimental-study-and-multivariate-evaluation 7/10

production   of   1-propanol   (descriptor  significance  100)  making

feasible  the  use  of   1-propanol   alone  as  a  quite  satisfactory

descriptor   for  calculating  ethanol   (Eq.   (1)),  although   it  was  the

second   most  abundant   produced   higher  alcohol.   However,  the

usage  of   three-descriptors  has  improved  the  model   and  has

introduced   1-butanol   (the  most  abundant)   and  even  isobutanol

(third   in  abundance)   as  relatively  important   determinants.  The

decision  on  which  model  could  be  the  most   applicable  depends   on

the  experimental  error  for  the  determination  of   each   descriptor   in

agreement  with  the  general  consideration  that  the  simplest  the

model   the  easier  its  application.

The  model   corresponding   to  E.  coli  for  predicting   ethanol

production   (Eq.  (5)) was  presumably  correlated   to  the  production

of   1-butanol   (descriptor  significance  100)  and  of   methyl-butanol(descriptor  significance  53.68).  Even  by  using  these  two  descrip-

tors   the  model   was  quite  satisfactory  (Eq.   (6)). The  use  of   three

descriptors  (1-butanol,   methyl-butanol   and  1-propanol)   resulted

practically  to  the  best  fit,  while  the  skip  of   isobutanol   as  a

descriptor   did  not  affect  the  results.

Finally,  the  model  corresponding   to  C.  sporogenes  appeared   to

be  the  most  complicated   one.   The  significant  deviation  of   the

experimental   values  from  the  relative  theoretical  ones   indicates

that  the  system  did  not  follow  the  linear  regression  but  a  more

complex   one.   Meanwhile   the  model  using   three  descriptors  (1-

propanol,   1-butanol   and  isobutanol)  appeared   to  be  the  most

satisfactory.

Our  contribution   introduces   the  alcohols   1-butanol,   1-propa-

nol, 

isobutanol, 

and 

methyl-butanol 

as 

potential 

biomarkers 

forquantifying  microbial  ethanol   production   and  not  only  as

qualitative  indicators   of   the  process.   This  is  the  main  difference

of   our  study   compared   to  previously  reported   results  by  other

groups.  The  study  of   Nanikawa  and  colleagues  has  correlated   the

ratio  of   1-propanol   to  ethanol   detected   in  rat  blood  and  skeletal

muscle   after  death  to  the  postmortem   formed  ethanol   [13].  Felby

and  Nielsen  in  their  study  have   correlated  the  blood  ethanol   and  1-

propanol   levels  to  the  state  of   putrefaction   of   the  body  [14], while

the  study  of   Moriya   and  Hashimoto  has  correlated  ethanol   to  1-

propanol   in  respect  to  the  stage  of   putrefaction  of   the  brain  of 

drowned   persons  [15].  Finally  1-butanol   has  been  suggested  by

Gubala   as  an  indicator   of   how  long   a  dead   body   has  been   laid  in

water  [16].  Our  results  are  in  agreement  with  the  aforementioned

studies 

in 

identifying 

1-propanol 

and 

1-butanol 

as 

main 

indicators

of   microbial  ethanol   production.   Furthermore,   our  results  identify

the  set  of   alcohols,  1-propanol,  1-butanol   and  isobutanol,  and  the

set  of   alcohols,  1-butanol,   1-propanol   and  methyl-butanol,  as

significant  biomarkers  of   ethanol   production   by  the  clostridial

species  (C.   perfrigens  and  C.  sporogenes)  and  E.  coli  respectively.

 3.5. 

 Applicability 

of  

the 

models 

in 

human 

blood 

and 

 plasma 

 products

The  applicability  of   the  constructed   models   was  tested  by

performing   anaerobic  cultures   of   normal   human  blood  and  plasma

inoculated   with  each   one  of   the  bacterial  species  at  25 

8C.  The

alcohols   content   was  determined  by  HS-GC–FID   at  days  0,  1,  2,  3,  5,

7,  11  and  15  and  the  experimental   results  were  compared   to  the

theoretical  results  deriving  from  the  relevant  models   for  C. perfigens,   C.  sporogenes  and  E.  coli, respectively.

The  following  blood  culture   systems  were  tested  in  parallel:

I.  whole  human  blood  with  EDTA  inoculated   with  C.    perfrigens;

II.  whole  human   blood  with  citric  ions   inoculated   with  C.

 perfrigens;

III.  human   plasma  inoculated  with  (A) 

E. 

coli; (B) 

C. 

sporogenes;  and

(C)  C.    perfrigens;

IV.  human   plasma  spiked  with  glucose  so  as  to  achieve  200  mg/dL 

initial  concentration   inoculated   with  (A) 

E. 

coli; and  (B) 

C.

sporogenes.

 3.5.1.   Alcohols   production

In 

the 

blood 

culture 

system 

‘‘I’’ 

ethanol 

and 

1-propanol 

wereproduced   during   the  fermentative  period,  while  in  the  presence   of 

citric  ions   (system  ‘‘II’’)  ethanol   was  produced   after  the  fifth  day  of 

fermentation,  without  production   of   detectable  amounts   of   any

other   alcohol.

When   human   plasma  was  used  as  the  culture   medium   and  was

inoculated   with 

E. 

coli  (IIIA)  and 

C. 

 perfrigens  (IIIC),  ethanol   and  1-

propanol   were   the  only   detectable  alcohols   produced   during

fermentation.  The  system  ‘‘IVA’’,  consisting  of   human   plasma

which   was  spiked  with  glucose  and  inoculated   with 

E. 

coli,

produced   higher  amounts   of   ethanol   than  the  system  ‘‘IIIA’’

consisting  of   human   plasma  inoculated   with  E.  coli  without

additional   glucose.  Interestingly,  the  ethanol   produced   in  the

system  ‘‘IVA’’  was  significantly  lower  than   the  ethanol   amount

produced 

in 

the 

E. 

coli 

cultures 

with 

BHI 

as 

the 

culture 

medium

Fig.  4.  (A)  Four  descriptors  model  and;  (B)  two  descriptor  model  for   ethanol  production  by  E.  coli.

V.A.   Boumba   et   al.  /   Forensic   Science  International   215  (2012)  189–198  195

Page 8: Microbial Ethanol Production Experimental Study and Multivariate Evaluation

7/18/2019 Microbial Ethanol Production Experimental Study and Multivariate Evaluation

http://slidepdf.com/reader/full/microbial-ethanol-production-experimental-study-and-multivariate-evaluation 8/10

(Table  4), although  the  initial  glucose  concentration   in both   culture

media  was  the  same  (200  mg/dL).  The  systems  ‘‘IIIB’’  and  ‘‘IVB’’

consisting   of   human  plasma  inoculated   with  C.  sporogenes  did  not

produce   ethanol   or  other  alcohols  during   the 15  days   of   incubation,

irrespectively  of   the  additional  glucose  (system  ‘‘IVB’’).

These  results   emphasize that  differences  in  the  composition  of 

the  culture  medium  (BHI,  whole  human  blood  and  human plasma)

can  affect  the  amount  of   the  produced  ethanol  by  each  microbe  and

they  are  in agreement  with  our  previous  observation  (Section  3.3).

The  cease  in  the  higher  alcohols production  in 

C. 

 pefrigens  human

blood  cultures  in  the  presence  of   citric  ions (system  ‘‘II’’)  may  be

attributed  to  the  flexibility   offered  by  the  aminoacids  fermentation

pathways (which generate the higher alcohols) which can be omitted

under  certain  conditions, as  suggested  earlier [5].  The determination

of   1-propanol  as  the  only   detectable higher  alcohol  produced  in the

microbial blood  and plasma  cultures  could be  attributed  to  its ability

to  be  produced  from many  different  substrates  and  fermentation

pathways,  in  response to  changing  environmental  conditions,  in

agreement  with   previous  suggestion  [5].

 3.5.2.  Testing   of   the  models

The  results  of   the  system  consisting  of   whole  human   blood

inoculated   with 

C. 

 perfrigens  (system  ‘‘I’’)  appear   to  be  very

complex   compared   to  the  relevant  model   as  this   was  described  by

Eq.  (1). The  experimental  values  cannot   be  predicted  successfully

by  PLS  regression  and  the  most  appropriate  fitting   model  is  a  2nd

degree  polynomial   (R2 =  0.79)  implying  the  difficulty  in  success-

fully  predicting   its  behavior.  An  alternative  methodology  might  be

to  remove   the  experimental  values  from  the  first  two  days,

although   our  intension  was  to  approach   the  volatiles  relationship

during   the  whole  experimental   time,  which   will  result  in  a  linear

relationship  with  R2 =  0.80.

In the  experiment   with  human   plasma  inoculated   with  C.

 perfrigens  (IIIC)   the  concentration   of   produced   1-propanol   remains

almost  constant   from   day  one  to  day  fifteen.  The  application  of   the

relevant  model  (Eq.  (1)) gives  a  maximum  error  of   22%  (average

16%).  The  model   also  confirms   the  linear  relationship  between

ethanol   and  1-propanol,   at  first  approximation,   showing 

R2 =  0.78

when   the  high   experimental  value  for  ethanol   measured  on  the

first  day  was  excluded.

The  model   of   E.   coli  has  a  quite  low  mean   square   error

(indicating  the  difference  between  the  true   values  of   the

concentrations   and  the  estimated  ones)  which  is  suggestive  of   a

successful  prediction.  The  results  have   showed  that   the  relevant

model   (Eq.   (5)) is  applicable  for  human  plasma  spiked with  glucose

(system  ‘‘IVA’’).  The  maximum   error  between  the  calculated   and

theoretical  values  of   ethanol   concentration   is  approximately  25%

(average  22%)   for  ‘‘IVA’’.  When   the  model   is  applied  to  the  system

 Table  5

Results  for  the  calculated  ethanol  concentrations  after  applying  each   model  (Eqs.  (1)–(6)) in  postmortem  cases  with   the  standard  error  (E%)  for  each   case.

#  Manner  of   death  Measured  ethanol  (g/L)   C.   perfrigens,

Eq.  (1)

C.   perfrigens,

Eq. (2)

E.  coli,   Eq.  (5)  E.  coli,  Eq.  (6)  C.  sporogenes,

Eq.  (3)

C.  sporogenes,

Eq.  (4)

Ethanol

(g/L)

E%  Ethanol

(g/L)

E%  Ethanol

(g/L)

E%  Ethanol

(g/L)

E%  Ethanol

(g/L)

E%  Ethanol

(g/L)

E%

1   Violent  0.10  0.13  26  0.05  151  0.24  128  0.15  44  0.24  131  0.27  156

2   Natural  0.16  0.15  8   0.05  129  0.71  344  0.80  399  0.24  53  0.29  82

3   Natural  0.16  0.09  43  0.07  145  0.20  26  0.15  8  0.18  8  0.21  28

4   Violent  0.19  0.22  18  0.00  99  0.32  69  0.15  20  0.43  128  0.44  136

5   Natural  0.20  0.06  70  0.09  146  0.18  12  0.15  25  0.11  45  0.15  27

6   Natural  0.21  0.30  45  0.05  78  0.39  85  0.15  29  0.58  176  0.59  179

Undetermined 

0.21 

0.20 

0.03 

116 

0.29 

38 

0.15 

29 

0.30 

41 

0.31 

488

 

Violent 

0.22 

0.27  23 

0.03 

87 

0.36  64 

0.15 

31 

0.52  139 

0.53  144

9   Undetermined  0.25  0.14  43  0.04  117  0.25  1  0.15  41  0.28  11  0.31  21

10 Undetermined  0.29  0.29  1   0.04  85  0.38  32  0.15  48  0.57  97  0.57  100

11  Undetermined  0.35  0.36  2   0.06  82  0.43  23  0.15  57  0.63  81  0.63  80

12 Natural  0.39  0.33  16  0.03  91  0.61  57  0.45  16  0.52  34  0.53  36

13  Natural  0.42  0.44  5   0.06  85  0.49  16  0.15  64  0.61  45  0.59  39

14  Violent  0.42  0.54  28  0.19  56  0.59  41  0.15  64  1.05  151  1.03  145

15  Natural  0.42  0.40  6   0.11  75  0.66  55  0.33  21  0.85  102  0.78  84

16  Undetermined  0.47  0.84  79  0.32  32  0.84  79  0.15  68  1.49  217  1.42  202

17 Natural  0.48  0.72  50  0.29  39  0.75  57  0.15  69  1.42  196  1.37  186

18 

Violent 

0.52 

0.65  26 

0.25 

51 

0.69  34 

0.15 

71 

1.28  149 

1.25  142

19  Undetermined  0.58  0.85  47  0.28  52  0.83  44  0.15  74  1.33  129  1.25  115

20  Undetermined  0.59  0.29  52  0.04  93  0.37  37  0.15  75  0.55  6  0.56  5

21 

Undetermined 

0.60 

0.63  6  

0.23 

61 

0.67  13 

0.15 

75 

1.21  104 

1.18  97

22  Undetermined  0.62  1.12  80  0.20  67  0.98  59  0.15  76  0.98  58  0.83  33

23  Violent  0.62  1.02  64  0.45  27  1.01  63  0.15  76  1.95  214  1.86  200

24  Undetermined  0.63  1.01  60  0.42  33  0.99  58  0.15  76  1.85  193  1.76  179

25  Undetermined  0.64  0.43  34  0.05  91  1.22  91  0.89  40  0.87  35  0.56  13

26  Undetermined  0.68  0.43  36  0.11  84  0.50  27  0.15  78  0.80  17  0.78  15

27  Undetermined  0.68  0.41  39  0.11  84  1.11  63  1.03  51  0.76  12  0.78  15

28  Violent  0.69  0.42  39  0.08  89  0.48  31  0.15  78  0.67  2  0.66  5

29  Undetermined  0.69  0.25  64  0.02  98  0.34  51  0.15  78  0.48  31  0.49  30

30  Undetermined  0.71  0.29  60  0.04  94  0.37  47  0.15  79  0.56  21  0.57  20

31 Natural 

0.72 

0.46 

35 

0.14 

81 

1.03  43 

0.82  13 

0.92  28 

0.88  23

32  Violent  0.73  0.67  8   0.26  64  0.71  3  0.15  79  1.30  78  1.26  73

33  Undetermined  0.75  1.00  34  0.46  38  1.00  33  0.15  80  2.00  166  1.91  155

34  Undetermined  0.81  0.30  64  0.03  96  0.74  9  0.63  22  0.56  32  0.54  33

35  Undetermined  0.87  0.95  9   0.42  52  1.56  80  0.99  13  1.84  112  1.76  103

36 Undetermined  0.91  0.92  2   0.42  54  0.93  3  0.15  83  1.84  103  1.76  95

37  Undetermined  0.94  1.13  20  0.33  65  1.72  83  0.94  0  1.61  71  1.32  40

38 Natural  0.98  0.43  56  0.10  90  0.49  50  0.15  85  0.74  24  0.73  26

39  Undetermined  1.08  0.29  73  0.04  96  0.37  65  0.15  86  0.56  49  0.56  48

40 Undetermined  2.58  1.08  58  0.49  81  1.06  59  0.15  94  2.08  19  1.99  23

The 

original 

ethanol 

concentrations 

measured 

for 

each 

case 

are 

also 

provided 

along 

with 

the 

manner 

of  

death.

V.A.  Boumba   et   al.  /   Forensic   Science  International   215  (2012)  189–198196

Page 9: Microbial Ethanol Production Experimental Study and Multivariate Evaluation

7/18/2019 Microbial Ethanol Production Experimental Study and Multivariate Evaluation

http://slidepdf.com/reader/full/microbial-ethanol-production-experimental-study-and-multivariate-evaluation 9/10

of   human   plasma  (IIIA)   the  model  loses  its  power  probably  due  to

the  fact  that  ethanol   and  1-propanol   values  are  strongly  correlated

exhibiting  a  steady  proportion   of   1:10.

Our  results  indicate   that  the  constructed   models   have  the

potential   to  be  applied  when   different  culture   media  are  used,

under  controlled  experimental  conditions.  Factors  such   as  the  type

of   microbial  species,  the  glucose  content   and  the  medium

composition  apparently   affect  the  procedure   of   microbial  ethanol,

1-butanol   and  higher   alcohols   production.

 3.6. 

Use 

of  

the 

models 

in 

real 

 postmortem 

cases

We  retrospectively  looked  our  chromatograms  archives  of   the

last   six  years  and  we  selected  40  chromatograms  corresponding  to

postmortem  cases, having  presence  of   other  alcohols  during  the

original  forensic ethanol  analysis.   The  classification  of these  cases  –

as  resulted  fromthe  retrospective review  ofthe  forensic  pathologist’s

reports  –  was:   10  casesof   natural  deaths;  8  cases of   violent  deaths;

and 22  casesof undetermined  cause  of  death.  Furthermore,  39  cases

had markedputrefactive phenomena at  autopsy  while  the  other  one

had  extensive  traumatic  lesions  (Case  no  4)  –  making  easier the

invasion  of microbes.  We calculated  for these  cases the  concentra-

tions of  higher alcohols and 1-butanol.  In Table 5  the original ethanol

concentrations measured  for each  case and  the  calculated  ethanolconcentrationsafter applying  each model  (Eqs.  (1)–(6)) are provided

along with  the  standard  error for each  case.

The  models  corresponding  to 

C. 

 perfrigens  estimated   the

microbial   produced  ethanol  with   a  standard  error  <40%  for  27

out of   the 40 cases  (68%),  26 of   them with  marked putrefaction  (96%).

Additionally,   21  of   these   cases   (78%)  had  ethanol  lower  than 0.7  g/L.

The  models  corresponding   to 

E. 

coli  estimated  the  microbial

produced   ethanol   with  a  standard  error  <40%  for  25  out  of   the  40

cases  (63%),  24  of   them  with  marked  putrefaction   (96%).  Moreover,

18  of   these  cases  (72%)  had  ethanol   lower  than   0.7  g/L.

The  models   corresponding   to  C.  sporogenes  estimated  the

microbial produced   ethanol   with  an  error  <40%  for 18  out  of   the 40

cases  (45%)  all  presented  with  putrefaction   at  autopsy,   while  12  of 

these  cases  (67%)  had  ethanol   lower  than   0.7  g/L.It  is  worth  mentioning   that   28  out  of   the  29  cases  (97%)  having

original  ethanol   concentration   lower  than   0.7  g/L   succeeded   a

standard   error 

<40%  in  predicting   the  microbially  produced

ethanol   by  at  least  one  of   the  models.

The caseswere  selected   due to the co-detection of   significant

amounts  of   volatiles  along  with   ethanol   during  the  original

ethanol analysis a  factor making the relevant  cases suspicious

for  microbial ethanol production, as  it   was   suggested [24]. Then

it  was  proved   that  the majority of   the selected cases had  marked

putrefaction at  autopsy   –  another  aggravating factor  for

microbial   activity   and ethanol  neo-formation   [14,15,24,25].

Our results  have shown  that  the models  could effectively  be

applied   in   postmortem   cases especially  when marked  putrefac-

tion 

is 

present. In addition, the models of  

C. 

 perfrigens 

showbetter applicability than these of   E.  

coli  and than those by C.

sporogenes.  Yet,   the  models are applied  more  satisfactory when

ethanol concentrations  are lower  than 0.7  g/L.   It is  generally

accepted  that  ethanol  concentrations   lower  than 0.7 g/L is  more

possible to be  of   microbial   origin for the  majority   of   cases with

neo-formation   of   ethanol  [2,3].

However,  it  has  to  be  underlined   that   so  far  the  study   of   real

postmortem   cases  meets  serious  limitations.  As  the  ethanol

concentrations   increase,  it  is  possible  part  of   the  detected   ethanol

to  be  of   ante  mortem   origin  making   the  proper   interpretation  of 

postmortem   ethanol   analysis  difficult.   In  such   a  case  it  is  extremely

challenging  and  premature   to  suggest  that   one  model   is  more

suitable  than   another.   On  the  other  hand,   other  microbes   might

generate 

different 

alcohols 

patterns 

resulting 

to 

different 

models.

This   might  to  explain  the  major  discrepancies  observed  between

measured   and  calculated  ethanol   concentrations   for  some  of   the

presented  cases.

Ideally, it  should  be known  which  microbial   species  have

been   present or  have been activated in  each   post-mortem case.

Practically,   more  mathematical   models, covering  a  large

spectrum of   bacterial or  fungal species should be available.

Moreover,   the identification of   more  influencing  factors and

the  quantification of   their  impact on  the  procedure  would

eventually   result to the construction of    more

accurate models  which could  be  applied   more  efficiently to

real cases.

Under  these views  it  is  naive to  suggest that such a  complicated

process,   with   so many  potential   effectors,   could  be entirely

covered   by the  simple models resulting  from  the  study of   only

three microbes.  However,   this   study represents  only an  initial

approach, albeit  with   some   useful   outcomes; it  is   promising   that

even   sucha  simplifiedapproach,highly challengingat  the present,

might provide a  methodical insight  into  this   huge, long  lasting

problem.

4.  Concluding   remarks

To  our  knowledge  the  present  contribution   is  a  first  approach   to

the  quantification  of   microbial  ethanol   production   by  demonstrat-

ing  a  mathematical   model   of   the  procedure   in  cases  where  other

alcohols   are  produced   simultaneously  with  ethanol.   The  microbial

biosynthesis  of   ethanol   in  parallel  with  the  biosynthesis  of   1-

butanol  and  the higher  alcohols  (1-propanol,   isobutanol, 2-methyl-

1-butanol   and  3-methyl-2-butanol)   identifies  these  alcohols   as

biochemical   biomarkers  suitable  for  quantifying  the  microbial

ethanol   production.

The  correlation  between  the  levels  of   the  alcohols   and  the

amount  of   the  microbially  produced   ethanol   is  expressed  by

mathematical   models  (equations).

The  alcohols   1-propanol   and  1-butanol   appear  to  have   a

preponderant 

role 

as 

descriptors 

of  

the 

relevant 

models.The  most  obvious  factor   affecting  the  produced   ethanol,   even

for  species  within  the  same  genus   (such  as  C.   perfrigens  and  C.

sporogenes),  is  the  microbe   type.

Differences  in  the  composition  of   the  culture   medium   (despite

the  initial  glucose  concentration)   result  in  the  production   of 

different  amounts   of   ethanol   and  other  alcohols   from  the  same

species.

The  overall  conclusion   of   the reported experimental study is

that  mathematical modeling  of   themicrobial ethanol  production

is   feasible, at  least  partially. Nevertheless,  given  the complexity

of   the  process,   extrapolation   of   the results  to  everyday practice

requires ongoing work  on  the identification and

quantification   of   the factors influencing the  postmortem   ethanol

production.

 Appendix    A.  Supplementary   data

Supplementary   data  associated  with   this   article  can  be  found,  in

the  online   version,   at  doi:10.1016/j.forsciint.2011.03.003 .

References

[1] C.L. O’Neal, A. Poklis, Postmortem production of ethanoland factors that influenceinterpretation, Am. J. Forensic Med. Pathol. 17 (1996) 8–20.

[2]  K. Ziavrou, V.A. Boumba, T. Vougiouklakis, Insights into theoriginof postmortemethanol,   Int. J. Toxicol. 24 (2005) 69–77.

[3] 

F.C. Kugelberg, A.W. Jones, Interpretingresults of ethanolanalysis in postmortem

specimens: a review of the literature, Forensic Sci. Int. 165 (2007) 10–29.

V.A.   Boumba   et   al.  /   Forensic   Science  International   215  (2012)  189–198  197

Page 10: Microbial Ethanol Production Experimental Study and Multivariate Evaluation

7/18/2019 Microbial Ethanol Production Experimental Study and Multivariate Evaluation

http://slidepdf.com/reader/full/microbial-ethanol-production-experimental-study-and-multivariate-evaluation 10/10

[4] J.E.L. Corry, Possible sources of ethanol ante- and postmortem: its relation to thebiochemistry andmicrobiologyof decomposition,J. Appl. Bacteriol. 44 (1978) 1–56.

[5]  V.A. Boumba, K. Ziavrou, T. Vougiouklakis, Biochemical pathways generatingpost-mortem volatile compounds co-detected during forensic ethanol analysis,Forensic Sci. Int. 174 (2008) 133–151.

[6] D. Yamima, H. Motani, K. Kamei, Y. Sato, M. Hayakawa, H. Iwase , Ethanolproduction by Candida albicans in post-mortem human blood samples: effectof   blood glucose level and dilution, Forensic Sci. Int. 164 (2006) 116–121.

[7]   J. Chang, S.E. Kollman, The effect of temperature on the formation of ethanol byCandida  albicans in blood, J. Forensic Sci. 34 (1989) 105–109.

[8]   J.J. Saady, A. Poklis, H.P. Dalton, Production of urinary ethanol after sample

collection, J. Forensic Sci. 38 (1993) 1467–1471.[9]  H.A. Sulkowski, A.H. Wu, Y.S. McCarter, In-vitro production of ethanol in urine byfermentation, J. Forensic Sci. 40 (1995) 990–993.

[10]  B.M. Appenzeller, M. Schuman, R. Wennig, Was a child poisoned by ethanol?Discriminationbetween ante-mortemconsumptionand post-mortemformation,Int.    J. Legal Med. 122 (2008) 429–434.

[11] A.C. Gruszecki, C.A. Robinson, S. Kloda, R.M. Brissie, High urine ethanol andnegative

 

blood and vitreous ethanol in a diabetic woman: a case report, retro-spective  case survey, and reviewof the literature, Am.J. ForensicMed. Pathol. 26(2005)  96–98.

[12] G.D. Amick, K.H. Habben, Inhibition of ethanol production by Saccharomycescerevisiae in human blood by sodiumfluoride, J. Forensic Sci. 42 (1997) 690–692.

[13] 

R.Nanikawa,K. Ameno, Y.Hashimoto,K. Hamada, Medicolegal studies on alcoholdetected in dead bodies – alcohol levels in skeletal muscle, Forensic Sci. Int. 20(1982)  133–140.

[14]  S. Felby, E. Nielsen, Congener production inbloodsamplesduring preparationandstorage,  Blutalkohol 32 (1995) 5–58.

[15] F.Moriya,Y. Hashimoto, Postmortemproductionof ethanolandn-propanol inthebrain  of drowned persons, Am. J. Forensic Med. Pathol. 25 (2004)131–133.

[16]  W. Gubala, n-Butanol in blood as the indicator of how long a dead body lay inwater,  Forensic Sci. Int. 46 (1990) 127–128.

[17]  C.L. Morris-Kukoski, E. Jagerdeo, J.E. Schaff, M.A. LeBeau, Ethanol analysis frombiological samples by dual rail robotic autosampler, J. Chromatogr. B Analyt.Technol.   Biomed. Life Sci. 850 (2007) 230–235.

[18]  D. Zuba, W. Piekoszewski, J. Pach, L. Winnik, A. Parczewski, Concentration of ethanolandother volatile compoundsin thebloodof acutelypoisoned alcoholics,Alcohol  26 (2002) 17–22.

[19] L. Kristoffersen, L.E.Stormyhr,A. Smith-Kielland,Headspace gas chromatographicdetermination of ethanol: the use of factorial design to study effects of bloodstorage  and headspace conditions on ethanol stability and acetaldehyde forma-tion  in whole blood and plasma, Forensic Sci. Int. 161 (2006) 151–1577.

[20]  K.J.Ryan, C.G.Ray (Eds.), SherrisMedicalMicrobiology,4th ed.,McGrawHill,2004.[21]  A.W. Jones,L. Hylen, E. Svensson,A. Helander,Storage of specimensat 4 degrees C

or  addition of sodium fluoride (1%) prevents formation of ethanol in urineinoculated with Candida albicans,   J. Anal. Toxicol. 23 (1999) 333–336.

[22]  E.R. Kuhn, A.K. Vickers, S.E. Ebleler, D.M. Ahlgren, J.H. Thorngate, Completeseparation and quantitation of fusel oils by capillary GC, Application, AgilentTechnologies, Inc. 2003.

[23] B.S.de Martinis, C.C.S. Martin, Automated headspace solid-phasemicroextractionand

 

capillarygas chromatography analysis of ethanolin post-mortemspecimens,Forensic  Sci. Int. 128 (2002) 115–119.

[24]  G. Scopp, Postmortem toxicology, Forensic Sci. Med. Pathol. 2 (2010) 314–325.[25]  D.M. Butzbach, Theinfluence of putrefactionand samplestorage on post-mortem

toxicology results, Forensic Sci. Med. Pathol. 6 (2010) 35–45.

V.A.  Boumba   et   al.  /   Forensic   Science  International   215  (2012)  189–198198