Content uploaded by Jesús Sánchez-Yagüe
Author content
All content in this area was uploaded by Jesús Sánchez-Yagüe
Content may be subject to copyright.
This article was downloaded by:[Sánchez-Yagüe, Jesús]
On: 29 April 2008
Access Details: [subscription number 792708647]
Publisher: Informa Healthcare
Informa Ltd Registered in England and Wales Registered Number: 1072954
Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Cancer Investigation
Publication details, including instructions for authors and subscription information:
http://www.informaworld.com/smpp/title~content=t713597231
Erythrocyte and Platelet Phospholipid Fatty Acids as
Markers of Advanced Non-Small Cell Lung Cancer:
Comparison with Serum Levels of Sialic Acid, TPS and
Cyfra 21-1
Javier de Castro a; Marina C. Rodríguez b; Vicenta S. Martínez-Zorzano c; Ángel
Hernández-Hernández d; Marcial Llanillo d; Jesús Sánchez-Yagüe d
aRadiology Service, Santísima Trinidad Foundation Hospital, Salamanca, Spain
bNeumosalud, Santísima Trinidad Foundation Hospital, Salamanca, Spain
cDepartment of Biochemistry, Genetics and Immunology, University of Vigo, Vigo,
Spain
dDepartment of Biochemistry and Molecular Biology, University of Salamanca,
Salamanca, Spain
Online Publication Date: 01 May 2008
To cite this Article: de Castro, Javier, Rodríguez, Marina C., Martínez-Zorzano, Vicenta S., Hernández-Hernández,
Ángel, Llanillo, Marcial and Sánchez-Yagüe, Jesús (2008) 'Erythrocyte and Platelet Phospholipid Fatty Acids as Markers
of Advanced Non-Small Cell Lung Cancer: Comparison with Serum Levels of Sialic Acid, TPS and Cyfra 21-1', Cancer
Investigation, 26:4, 407 - 418
To link to this article: DOI: 10.1080/07357900701788114
URL: http://dx.doi.org/10.1080/07357900701788114
PLEASE SCROLL DOWN FOR ARTICLE
Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf
This article maybe used for research, teaching and private study purposes. Any substantial or systematic reproduction,
re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly
forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents will be
complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be
independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings,
demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or
arising out of the use of this material.
Downloaded By: [Sánchez-Yagüe, Jesús] At: 11:16 29 April 2008
Cancer Investigation, 26:407–418, 2008
ISSN: 0735-7907 print / 1532-4192 online
Copyright c
Informa Healthcare USA, Inc.
DOI: 10.1080/07357900701788114
ORIGINAL ARTICLE
Imaging, Diagnosis, Prognosis
Erythrocyte and Platelet Phospholipid Fatty Acids as
Markers of Advanced Non-Small Cell Lung Cancer:
Comparison with Serum Levels of Sialic Acid, TPS
and Cyfra 21-1
Javier de Castro,1Marina C. Rodr´ıguez,2Vicenta S. Mart´ınez-Zorzano,3´
Angel Hern ´andez-Hern´andez,4
Marcial Llanillo,4and Jes ´us S ´anchez-Yag¨ue4
Radiology Service,1Neumosalud,2Sant´ısima Trinidad Foundation Hospital, Salamanca, Spain.
Department of Biochemistry, Genetics and Immunology, University of Vigo, Vigo, Spain.3
Department of Biochemistry and Molecular Biology, University of Salamanca, Salamanca, Spain.4
ABSTRACT
The phospholipid fatty acid profiles of erythrocytes and platelets from fifty patients with
advanced non-small cell lung cancer were investigated using gas chromatography/mass spec-
trometry, followed by “ROC” curves analysis to gain novel biomarker information. Sialic acid
and cytokeratins were also examined. Potentially useful fatty acid markers: Erythrocytes: phos-
phatidylcholine, 18:2n6 and 20:4n6; phosphatidylethanolamine, 22:4n6 and 22:6n3 +24:1n9.
Platelets: phosphatidylcholine, 22.0; phosphatidylethanolamine, 22:5n3 +24:0. At the cut-
off value to obtain maximum accuracy, the best biomarkers were found in platelets: phos-
phatidylserine +phosphatidylinositol (PS +PI), 21:0; sphyngomyelin: 20:1n9 and 22:1n9. All
these fatty acids showed similar/higher diagnostic yields than the commonly used markers
sialic acid or cytokeratins.
INTRODUCTION
There is growing interest in finding new substances or pa-
rameters that could be used as tumor markers for neoplastic
diseases in order to gain more knowledge about these diseases
and improve therapeutic approaches.
Keywords: Tumor markers, Non-small cell lung cancer, Fatty acids,
Sialic acids, Cytokeratins, Erythrocyte, Platelet.
This work has been supported in part by FIS PI020081 and the
Junta de Castilla y Le´on (SA 126A07 and Biomedicina,
SAN191/SA29/06).
Correspondence to:
Jes´us S ´anchez Yag¨ue PhD
Department of Biochemistry and Molecular Biology, University of
Salamanca
Edificio Departamental. Lab. 106
Plaza Doctores de la Reina s/n
37007 Salamanca
Spain
email: sanyaj@usal.es
Different laboratory studies have shown increased sialic acid
levels at the surface of malignant or transformed cells (1, 2).
Since a rapid turnover of sialic acid from cell components ap-
pears in the circulation, elevations in sialic acid levels in serum
have been used as a tumor marker for the diagnosis and man-
agement of cancer patients (3) specifically lung cancer patients
(4–6).
Cytokeratins are intermediate filaments expressed by epithe-
lial cells and their malignant counterparts (7). Some cytokeratin
fragments are released to the serum owing to cell lysis or tu-
mor necrosis, and hence serum cytokeratins, mainly TPS (tissue
polypeptide-specific antigen) and Cyfra 21-1 (cytokeratin frag-
ment recognized by the KS 19-1 and BM 19-21 antibodies) have
been used as markers of cancer, including lung cancer (8).
Furthermore, metabolic profiling has increasingly been used
as a probe in disease diagnosis and pharmacological analysis (9).
Essential fatty acids play an important role in complex metabolic
reactions, and essential polyunsaturated fatty acids (PUFA) ap-
pear to be one of the critical targets in the complex metabolic
stages that lead to, or are associated with, cancer. Although ev-
idence of the alterations in lipid and fatty acid metabolism in
407
Downloaded By: [Sánchez-Yagüe, Jesús] At: 11:16 29 April 2008
cancer patients is well documented (10–13), the potential use
of fatty acid profiles, especially from erythrocytes or platelets,
as markers in clinical oncology has not been addressed, and
only recently has plasma fatty acid metabolic profiling assessed
by gas chromatography/mass spectrometry (GC/MS) been used
to detect biomarkers of type 2 diabetes mellitus (14). In this
regard, we have recently described that the fatty acid compo-
sition of total lipids from erythrocytes and platelets is altered
in patients with advanced non-small cell lung cancer (NSCLC),
probably as a result of the formation of free radicals and lipid
peroxidation processes (15). The latter two phenomena are both
linked both to carcinogenesis and tumor behavior (16, 17).
Accordingly, in this study we analyzed the fatty acid profiles
of phospholipid species in erythrocytes and platelets from pa-
tients with advanced NSCLC to check which of those fatty
acids might eventually provide an additional test for the diag-
nosis and/or management of NSCLC patients. For comparison,
we also examined previously described markers for lung car-
cinoma such as sialic acid (FSA, free sialic acid; TSA, total
sialic acid; BSA, glicoconjugated-bound sialic acid; TSA/TP,
total sialic acid/total protein; BSA/TP, glycoconjugated-bound
sialic acid/total protein) and the cytokeratins TPS and
Cyfra 21–1.
MATERIALS AND METHODS
Patients
We studied 50 patients with a clear histological diagno-
sis of primary NSCLC (39 squamous cell carcinoma, and 11
adenocarcinoma) according to the World Health Organization
(WHO) classification. Eligibility criteria included: no previous
chemotherapy or radiotherapy; ECOG performance status ≤2;
adequate bone marrow, liver, renal and cardiac function; no
known brain metastasis; no previous malignancy; no serious
concurrent medical illness, and no extreme dietary habits. The
staging of the tumors, performed according to the TNM stag-
ing system of the International Union Against Cancer (UICC),
was IIIA, IIIB and IV for 9, 28 and 13 patients, respectively.
For TNM staging, all patients underwent a computed tomogra-
phy (CT) scan of the chest and upper part of the abdomen, a
bone scintigram, and a brain CT or magnetic resonance imag-
ing. These subjects were well informed as to the purpose of, and
fully agreed to participate in the study, which was performed
after approval by our local institutional review boards at the
University of Salamanca and the Sant´ısima Trinidad Founda-
tion Hospital in Salamanca. We also studied 50 healthy male and
female volunteers as controls, whose age, body weight, blood
lipids, blood pressure, smoking habits and BMI (body mass in-
dex) were equivalent to those of the patient group (Table 1).
Blood samples were taken the day after the patients were in-
formed of their illness. None of the individuals had received
drugs interfering with platelet or erythrocyte functions, clotting,
or fibrinolytic activity within the three months preceding blood
sampling.
Tab l e 1. Clinical and biochemical characteristics of control subjects
and NSCLC patients
Lung cancer
Var iable Controls patients
Age (years) 61 ±568±5
Weight (kg) 67 ±10 69 ±12
Mean arterial pressure (mm Hg) 91 ±10 95 ±5
Platelets (x 103mL) 191 ±54 261 ±69
Erythrocytes (x 106mL) 5.2 ±0.9 4.7 ±0.9
Total cholesterol (mg/dL) 180 ±47 177 ±25
HDL-cholesterol (mg/dL) 52 ±10 53 ±10
Tr iacylglycerols (mg/dL) 83 ±30 89 ±27
Smoking history (pack-years) >19 ≥20
BMI (kg/m2) 25.4 ±4.2 25.0 ±2.8
Values given as means ±S.D. BMI, body mass index.
Blood sampling and isolation of serum,
plasma, erythrocyte and platelets
Two blood samples were drawn from all the individuals.
One sample was allowed to coagulate at room temperature and
serum was then obtained after centrifugation. The other one cor-
responded to heparinized blood samples. The latter were cen-
trifuged at 1,000 ×gfor 5 minutes to obtain a lower phase (ery-
throcytes) and an upper phase (platelet-rich plasma). Two thirds
of the platelet-rich plasma was then centrifuged at 2,600 ×gfor
10 minutes. The platelet pellet was washed three times with 3
volumes of a solution of 150 mM NaCl, 5 mM sodium phosphate
buffer, pH 8.0 (PBS). After washing, platelets were resuspended
in 10 mM HEPES, pH 6.5, 0.2 mM EGTA, 5 mM KCl, 5.5 mM
glucose in order to obtain platelet homogenates (18,19). Isolated
erythrocytes were washed as indicated above for platelets. The
serum or plasma was stored at −80◦C until tested.
Biochemical measurements in serum
and plasma
β-Thromboglobulin (β-TG) and Platelet Factor 4 (PF4)
were measured in plasma by the ELISA technique (Boehringer
Mannheim Italy, Milan, Italy) following the manufacturer’s
instructions.
Sialic acids and cytokeratins were measured in serum. FSA
was determined by means of the thiobarbituric acid method (20)
according to the procedure described by Aminoff (21). TSA
was quantified by the same procedure after hydrolysis of the
samples in five volumes of 0.1N H2SO4at 80◦C for one hour.
BSA was calculated as the difference between TSA and FSA.
The levels of Cyfra 21-1 were determined using a double de-
terminant immunoradiometric assay (ELSA Cyfra 21-1; CIS
Biointernational; Gif-sur-Yvette; Cedex, France). Two different
monoclonal antobodies (KS 19.1 catcher and BM 19.21 tracer
antibodies) were used in this two-step sandwich assay. TPS was
measured using a two-site immunoradiometric assay, employing
two monoclonal antibodies to the M3 epitope of TPA, the former
obtained from mice and the latter from horses (TPS IRMA; Beki
Diagnostics AB; Bromma, Sweden). Serum protein contents
408 J. de Castro et al.
Downloaded By: [Sánchez-Yagüe, Jesús] At: 11:16 29 April 2008
(TP) were estimated by the Biuret method (22), using bovine
serum albumin as standard.
Lipid Extraction, Determination of
Phospholipid Classes and Fatty Acid Analyses
Total lipids were extracted from erythrocytes or platelets
with mixtures of isopropanol/chloroform (23). Separation of
the phospholipid classes was accomplished by two-dimensional
thin layer chromatography (24). Fatty acid analyses were car-
ried out by gas chromatography-mass spectrometry (GC-MS)
(15, 24, 25). Before analysis, wet silica gel areas containing
the individual phospholipids were transmethylated with BF3
reagent, and fatty acid methyl esters (FAME) were partitioned
in a water/petroleum hydrocarbon system, brought to dryness
under nitrogen, and dissolved in 10 µLofisooctane (24). GC-
MS (triplicate determinations per sample) was performed on a
Shimadzu 17A apparatus with a TR-Wax (30 m ×0.25 mm ×
0.25 µm) capillary column connected to a Shimadzu MS QP
5000. GC was programmed under the following temperature
gradient: initial time at 130◦C for 3 minutes, 15◦C/minute from
130◦Cto190
◦C, 2◦C/minute from 190◦Cto230
◦C and holding
at 230◦Cfor 23 minutes, with helium as carrier gas. Injector and
detector temperatures were 210◦C and 240◦C, respectively.
Statistical analyses
Data were analyzed using the non-parametric Mann-Whitney
Utest. Statistical significance was considered at P<0.05.
“ROC” (receiver operating characteristics) curves, which cor-
relate the percentages between true and false positives, were
calculated in order to select the cut-off level. The cut-off point
for each tumor marker was selected on the basis of the best ac-
curacy. If several cut-off points had the same accuracy, the value
with the best specificity was chosen. Correlation indices were
assessed using Spearman’s correlation coefficient test. Analyses
were implemented using the SPSS program for MS Windows
(version 13.0.1).
RESULTS
Levels of TP and biomarkers in serum
The serum levels of all sialic acid and cytokeratin biomarkers
were significantly increased in the NSCLC patients (Table 2). No
significant differences in TP values were found between patients
and controls (83.4 ±8.1 vs. 83.3 ±8.4 mg/mL in controls and
NSCLC patients, respectively).
Profiles of phospholipid fatty acids in
erythrocytes and platelets
The β-TG/PF4 ratio was not <2.5 in any of the samples,
and hence incorrect sampling and manipulation of the blood
withdrawn was ruled out (26).
The fatty acid profile of individual phospholipids from ery-
throcytes changed significantly in the lung cancer patients
(Tables 3–7). In phosphatidylcholine (PC), the most significant
changes observed were the decreases in 18:2n6, 20:4n6, 20:5n3
Tab l e 2. Serum levels of the biomarkers sialic acid and the cytokeratins
TPS and Cyfra 21-1 in control subjects and NSCLC patients
NSCLC Statistical
Controls patients significance
FSA (µmol/mL) 0.027 ±0.008 0.031 ±0.007 P<0.033
TSA (µmol/mL) 2.25 ±1.04 2.89 ±0.67 P<0.001
BSA (µmol/mL) 2.22 ±1.04 2.86 ±0.67 P<0.001
TSA/TP (µmol/mg) 0.027 ±0.011 0.035 ±0.008 P<0.001
BSA/TP (µmol/mg) 0.026 ±0.011 0.034 ±0.008 P<0.001
TPS (U/L) 45.14 ±79.3 96.7 ±58.2 P<0.005
Cyfra 21.1 (ng/mL) 0.98 ±0.49 5.99 ±9.15 P<0.044
TPS/TP 0.54 ±0.48 1.15 ±0.69 P<0.005
Cyfra 21.1/TP 0.014 ±0.009 0.07 ±0.103 P<0.045
Values given as means ±S.D. FSA, free sialic acid; TSA, total sialic
acid; BSA, glycoconjugate-bound sialic acid; TP, total protein; TPS,
tissue polypeptide specific antigen; Cyfra 21-1, cytokeratin fragment
recognized by KS 19-1 and BM 19-21 antibodies.
and 22:6n3 +24:1n9 (23, 50, 68 and 39%, P<0.001, respec-
tively). By contrast, 18:1n9 increased significantly (10%, P<
0.015). These changes mainly contributed to the significant de-
crease in polyunsaturated fatty acids (PUFA) (15%, P<0.006),
and they elicited a significant reduction in the unsaturation in-
dex (UI) (10%, P<0.025). In phosphatidylethanolamine (PE),
20:4n6, 20:5n3 and 22:6n3 +24:1n9 decreased significantly
(39%, P<0.001; 41%, P<0.003; 47%, P<0.001, respec-
tively), whereas 16:0, 18:0 and 18:1n9 increased significantly
(10%, P<0.009; 17%, P<0.031; 10%, P<0.011, re-
spectively). These changes mainly contributed to the significant
decrease in PUFA and total unsaturated fatty acids (UFA) (24%,
P<0.004; 6%, P<0.030, respectively). The UI was also sig-
nificantly reduced (18%, P<0.004). In phosphatidylserine +
phosphatidylinositol (PS +PI), 16:0 and 18:1n9 decreased sig-
nificantly (33%, P<0.001; 26%, P<0.003, respectively),
whereas 18:0 increased significantly (13%, P<0.037). In this
case, monounsaturated fatty acids (MUFA) decreased signifi-
cantly (25%, P<0.041), although the fatty acid changes did
not elicit any variations in the UI. The fatty acid changes in
sphyngomyelin (SM) were less numerous, with significant de-
creases in 18:1n9 and 22:0 (37%, P<0.024; 18%, P<0.035,
respectively). Finally, in phosphatidic acid (PA), among the main
fatty acids, 18:2n6 decreased significantly (35%, P<0.013),
whereas 18:1n9 increased significantly (14%, P<0.002).
These changes mainly contributed to the significant decrease
in PUFA (33%, P<0.006) and in the UI (17%, P<0.021).
The fatty acid profile of individual phospholipids from
platelets also changed significantly in the lung cancer patients
(Tables 3–7). In PC, decreases occurred in 20:0, 20:1n9, 20:5n3
and 22:0 (41%, P<0.041; 42%, P<0.037; 56%, P<
0.013; and 71%, P<0.05, respectively). By contrast, 16:0
and 18:1n9 increased significantly (20%, P<0.034 and P<
0.008, respectively). These changes contributed to the significant
increase in saturated fatty acids (SFA) (15%, P<0.040). In PE,
20:0, 22:1n9 and 22:5n3 +24:0 decreased significantly (40%,
P<0.006; 58%, P<0.009; 58%, P<0.018, respectively),
Fatty Acids as Markers of Lung Cancer 409
Downloaded By: [Sánchez-Yagüe, Jesús] At: 11:16 29 April 2008
Tab l e 3. Changes in the fatty acid composition of phosphatidylcholine from erythrocytes and platelets in control subjects and NSCLC patients
Erythrocytes Platelets
Fatty acid Controls NSCLC patients Statistical significance Controls NSCLC patients Statistical significance
14:0 0.74 ±0.43 0.75 ±0.30 NS 0.95 ±0.34 1.82 ±1.38 N.S.
15:0 0.26 ±0.17 0.39 ±0.10 P<0.001 0.50 ±0.21 0.67 ±0.62 N.S.
16:0 37.84 ±6.52 37.66 ±6.82 N.S. 24.26 ±6.14 29.18 ±6.28 P<0.034
16:1n7 0.97 ±0.57 1.24 ±0.61 N.S. 2.06 ±0.89 1.78 ±0.52 N.S.
17:0 0.55 ±0.21 0.74 ±0.18 P<0.001 0.63 ±0.30 0.76 ±0.38 N.S.
17:1 0.21 ±0.15 0.28 ±0.21 N.S. 0.43 ±0.21 0.36 ±0.20 N.S.
18:0 14.50 ±2.19 15.28 ±3.04 N.S. 14.74 ±3.23 16.39 ±2.85 N.S.
18:1n9 20.99 ±3.48 22.98 ±3.37 P<0.015 20.97 ±4.18 25.28 ±3.51 P<0.008
18:2n6 20.46 ±3.18 15.57 ±5.83 P<0.001 17.99 ±9.21 11.80 ±3.37 N.S.
20:0 0.28 ±0.26 0.36 ±0.39 N.S. 1.00 ±0.58 0.59 ±0.32 P<0.041
20:1n9 0.68 ±0.93 0.74 ±0.61 N.S. 1.68 ±0.73 0.98 ±0.36 P<0.037
20:2n6 0.19 ±0.31 0.36 ±0.49 N.S. N.D. N.D.
20:4n6 2.88 ±1.02 1.40 ±1.03 P<0.001 5.07 ±1.34 4.73 ±2.03 N.S.
20:5n3 0.29 ±0.24 0.09 ±0.11 P<0.001 0.57 ±0.24 0.25 ±0.09 P<0.013
21:0 1.12 ±0.66 0.76 ±0.40 P<0.026 1.26 ±0.63 0.98 ±0.45 N.S.
22:0 0.18 ±0.23 0.23 ±0.30 N.S. 1.38 ±0.67 0.40 ±0.23 P<0.05
22:1n9 0.45 ±0.39 0.33 ±0.47 N.S. 1.11 ±0.77 0.56 ±0.45 N.S.
22:2 0.07 ±0.27 0.23 ±0.31 P<0.001 N.D. N.D.
22:4n6 0.15 ±0.20 0.07 ±0.20 P<0.003 N.D. N.D.
22:5n3 +24:0 0.52 ±0.36 0.41 ±0.60 P<0.038 1.23 ±0.46 0.94 ±0.39 N.S.
22:6n3 +24:1n9 1.15 ±0.61 0.70 ±0.58 P<0.001 N.D. N.D.
SFA 55.60 ±7.45 56.26 ±6.24 N.S. 43.47 ±7.44 49.82 ±5.22 P<0.040
MUFA 22.29 ±5.80 25.03 ±5.83 P<0.015 26.64 ±3.98 30.47 ±5.54 N.S.
PUFA 21.97 ±9.34 18.59 ±6.36 P<0.006 24.52 ±10.96 17.88 ±4.25 N.S.
TOTAL UFA 44.26 ±7.61 43.62 ±6.24 N.S. 51.12 ±10.84 47.29 ±4.93 N.S.
SFA/UFA 1.34 ±0.49 1.34 ±0.35 N.S. 0.96 ±0.36 1.02 ±0.18 N.S.
UI 76.84 ±17.98 68.72 ±12.98 P<0.025 89.72 ±24.44 77.25 ±13.23 N.S.
Data are given in percentages of total fatty acid contents and are means ±S.D. with GLC determinations performed in triplicate. N.D., not detected;
N.S., not significant; MUFA, monounsaturated fatty acids; PUFA, polyunsaturated fatty acids; SFA, saturated fatty acids; UFA, unsaturated fatty
acids; UI, unsaturation index, calculated as the sum of the percentage by weight of each fatty acid times the number of olefinic bonds.
whereas mainly 18:0 and 20:4n6 increased significantly (64%,
P<0.017; 26%, P<0.033; 154%, P<0.025, respectively),
although the changes did not elicit variations in the saturation or
unsaturation indices. In PS +PI, decreases in 21:0 and 22:1n9
(63%, P<0.001; 37%, P<0.041, respectively) were also
insufficient to elicit variations in the saturation or unsaturation
indices. Finally, in SM, the most significant changes were the
decreases in 18:2n6, 20:1n9, and 22:1n9 (40%, P<0.015;
70%, P<0.001; and 62%, P<0.008, respectively), whereas
16:0 increased significantly (70%, P<0.001). These changes
mainly contributed to the significant increase in SFA (30%, P<
0.002) and the decrease in the UI (43%, P<0.009).
A detailed analysis of the changes in the fatty acids also re-
vealed that some of those were cell specific while others were
common to both erythrocyte and platelets. The most signifi-
cant specific changes were: PC, decreases in 18:2n6, 20:4n6,
22:5n3 +24:0 and 22:6n3 +24:1n9 in erythrocytes, and an in-
crease in 16:0 in platelets. PE, decreases in 20:4n6, 20:5n3 and
22:6n3 +24:1n9 in erythrocytes, and a decrease in 22:5n3 +
24:0 in platelets. PS +PI, decreases in 16:0, 18:1n9 and
22:5n3 +24:0 and an increase in 18:0 in erythrocytes. SM,
a decrease in 22:0 in erythrocytes, and an increase in 16:0 in
platelets. The most significant common changes were: PC, a de-
crease in 20:5n3 and an increase in 18:1n9; PE, an increase in
18:0. We did not detect common changes in PS +PI or SM fatty
acids.
ROC curves for biomarkers in serum
and phospholipid fatty acids in erythrocytes
and platelets
We then calculated ROC curves for most of the erythrocyte
or platelet phospholipid fatty acids that changed in the NSCLC
patients, as well as for the tumor markers (Table 8) in order
to compare the different markers by analyzing their diagnostic
accuracy (area below the ROC curve). We observed that the di-
agnostic accuracy of several phospholipid fatty acids was similar
to or greater than that of sialic acid (0.81 for TSA/TP and 0,79
for TSA and BSA) or cytokeratins (0.76 and 0.68 for TPS and
Cyfra 21-1, respectively). The most significant were: Erythro-
cytes: PC, 18:2n6 (0.81) and 20:4n6 (0.85); PE, 22:4n6 (0.84),
22:6n3 +24:1n9 (0.82); PA, 20:5n3 (0.93). Platelets: PC, 18:1n9
(0.81), 20:5n3 (0.93), 22:0 (0.96); PE, 20:0 (0.88), 20:4n6 (0.80),
22:1n9 (0.92), 22:5n3 +24:0 (0.93); PS +PI, 21:0 (0.97), SM,
16:0 (0.91), 20:1n9 (0.96), 22:1n9 (0.97).
The operating characteristics for the individual fatty acids and
tumor markers with their cut-off points for achieving the best in-
dividual accuracy are shown in Table 9. The highest specificity,
410 J. de Castro et al.
Downloaded By: [Sánchez-Yagüe, Jesús] At: 11:16 29 April 2008
Tab l e 4. Changes in the fatty acid composition of phosphatidylethanolamine from erythrocytes and platelets in control subjects and NSCLC patients
Erythrocytes Platelets
Fatty acid Controls NSCLC patients Statistical significance Controls NSCLC patients Statistical significance
14:0 1.09 ±0.73 0.95 ±0.59 N.S. 2.14 ±0.87 2.23 ±0.68 N.S.
15:0 0.33 ±0.24 0.34 ±0.36 N.S. 0.74 ±0.35 0.56 ±0.25 N.S.
16:0 24.46 ±3.09 26.77 ±3.21 P<0.009 15.85 ±5.34 15.18 ±4.23 N.S.
16:1n7 0.95 ±0.50 1.02 ±0.62 N.S. 2.74 ±0.91 2.46 ±0.87 N.S.
17:0 0.58 ±0.21 1.03 ±0.76 P<0.002 1.25 ±0.56 1.02 ±0.53 N.S.
17:1 1.09 ±0.71 1.31 ±1.12 N.S. 0.70 ±0.46 1.65 ±1.14 P<0.017
18:0 13.91 ±2.43 16.34 ±4.79 P<0.031 18.21 ±6.17 23.00 ±3.68 P<0.033
18:1n9 28.01 ±2.95 30.74 ±4.43 P<0.011 17.89 ±7.84 17.81 ±6.69 N.S.
18:2n6 9.36 ±3.72 7.96 ±3.38 N.S. 7.54 ±3.99 8.68 ±4.27 N.S.
20:0 0.23 ±0.46 0.43 ±0.56 P<0.007 1.65 ±0.42 0.99 ±0.42 P<0.006
20:1n9 0.50 ±0.33 0.77 ±0.59 N.S. 1.90 ±0.73 1.17 ±0.85 N.S.
20:4n6 10.92 ±3.56 6.67 ±4.40 P<0.001 5.94 ±3.67 15.12 ±9.16 P<0.025
20:5n3 0.68 ±0.52 0.40 ±0.56 P<0.003 0.67 ±0.37 0.49 ±0.16 N.S.
21:0 0.77 ±0.33 0.77 ±0.49 N.S. 0.86 ±0.67 0.44 ±0.15 N.S.
22:0 0.46 ±0.56 0.29 ±0.40 N.S. 1.07 ±0.98 0.57 ±0.28 N.S.
22:1n9 0.41 ±0.29 0.34 ±0.42 N.S. 1.10 ±0.47 0.46 ±0.26 P<0.009
22:4n6 3.52 ±1.04 1.67 ±1.57 P<0.001 2.65 ±1.34 2.51 ±1.04 N.S.
22:5n3 +24:0 0.55 ±1.14 1.11 ±1.49 P<0.033 2.27 ±0.84 0.94 ±0.63 P<0.018
22:6n3 +24:1n9 2.97 ±1.21 1.57 ±1.29 P<0.001 N.D. N.D.
SFA 41.94 ±5.75 44.14 ±9.91 N.S. 43.12 ±10.82 45.16 ±8.03 N.S.
MUFA 31.75 ±2.86 34.59 ±4.55 P<0.012 24.52 ±9.10 22.76 ±6.88 N.S.
PUFA 24.85 ±9.42 18.75 ±6.94 P<0.004 26.09 ±14.07 32.88 ±11.91 N.S.
TOTAL UFA 56.60 ±8.08 53.34 ±7.08 P<0.030 52.11 ±14.72 53.60 ±10.24 N.S.
SFA/UFA 0.77 ±0.23 0.84 ±0.27 N.S. 0.98 ±0.64 0.91 ±0.41 N.S.
UI 118.05 ±28.52 96.25 ±25.04 P<0.004 98.09 ±36.39 120.12 ±46.96 N.S.
Data are given in percentages of total fatty acid contents and are means ±S.D. with GLC determinations performed in triplicate. N.D., not detected;
N.S., not significant; MUFA, monounsaturated fatty acids; PUFA, polyunsaturated fatty acids; SFA, saturated fatty acids; UFA, unsaturated fatty
acids; UI, unsaturation index, calculated as the sum of the percentage by weight of each fatty acid times the number of olefinic bonds.
sensitivity, positive predictive value (PPV) and accuracy were
found with: (a) Erythrocytes: PC fatty acids, 20:5n3 and 18:2n6
(90 and 68%, 56 and 81%, 83 and 78%, and 74 and 75%, re-
spectively); PE fatty acids, 20:4n6 and 22:4n6 (77 and 65%,
71 and 75%, 85 and 77%, and 73 and 71%, respectively); PS
+PI fatty acids, 16:0 and 22:5n3 +24:0 (74 and 100%, 89
and 24%, 85 and 100%, and 84 and 63%, respectively); SM
fatty acids, 22:0 (94, 64, 96 and 61 %, respectively); PA fatty
acids, 18:1n9, 22:5n3 and 22:1n9 (67, 80 and 80%; 81, 57 and
50%; 78, 57 and 57%, and 75, 73 and 69%, respectively); (b)
Platelets: PC fatty acids, 20:5n3 and 22:0 (80 and 88%, 100
and 83%, 86 and 83%, and 92 and 86%, respectively); PE fatty
acids, 22:1n9 and 22:5n3 +24:0 (86 and 80%, 86 and 100%,
86%, and 86 and 91%, respectively); PS +PI fatty acids, 21:0
(80, 91, 83 and 86%, respectively); SM fatty acids, 16:0 and
18:2n6 (77 and 75%, 82 and 89%, 75 and 73%, and 79 and
81%, respectively). As regards serum markers, with respect to
the sialic acids BSA showed the best accuracy, specificity and
PPV (84, 83 and 92%), while FSA showed the best sensitiv-
ity (95%). Additionally, when sialic acids were normalized by
TP, the accuracy, specificity and PPV values decreased, while
the sensitivity values remained similar (74%, 44%, 79% and
88 and 86% for TSA/TP and BSA/TP, respectively). We also
found positive correlations among TSA, FSA, BSA, FSA/TP
and BSA/TP (data not shown); e.g., between TSA and BSA or
FSA and BSA, we found Spearman’s correlation coefficients
of 1 (P<0.0001) and 0.886 (P<0.0001) or 0.303 (P<
0.017) and 0.272 (P<0.034) when BSA values were not
normalized or normalized by TP, respectively. Both cytoker-
atins showed similar values, although TPS proved to be slightly
superior at the cut-off point for achieving the best individual
accuracy (37 U/L and 1 ng/mL, for TPS and Cyfra 21-1, re-
spectively). Nevertheless, at commonly reported cut-off val-
ues (140 U/L and 3.6 ng/mL for TPS and Cyfra 21-1, respec-
tively, both with 100% specificity), Cyfra 21-1 showed other
better values than TPS (39, 100, 56% and 24, 100, 45% sen-
sitivity, PPV and accuracy, respectively. Data not shown). It
has been reported that the correlation index between the cy-
tokeratins TPA and Cyfra 21-1 is very high [27]. Here, we
found a positive correlation between the two cytokeratin mark-
ers TPS and Cyfra 21-1 [Spearman’s correlation coefficients
of 0.651 and 0.637 (P<0.001) when the markers were nor-
malized or not normalized by TP, respectively]. No correlation
was found between either of the cytokeratins and sialic acid
values.
According to the data shown in Table 9, when the diagnos-
tic yields (sensitivity and specificity) were analyzed at the cut
off point to achieve the best accuracy, some fatty acids showed
similar or even higher values than cytokeratins. Especially sig-
nificant were: erythrocytes: PC, 18:2n6 and 20:4n6 (81 and 68%,
and 79 and 65%, respectively); PE, 20:4n6, 22:4n6 and 22:6n3
+24:1n9 (71 and 77%, 75 and 65%, and 71%, respectively);
Fatty Acids as Markers of Lung Cancer 411
Downloaded By: [Sánchez-Yagüe, Jesús] At: 11:16 29 April 2008
Tab l e 5. Changes in the fatty acid composition of phosphatidylserine +phosphatidylinositol from erythrocytes and platelets in control subjects and
NSCLC patients
Erythcocytes Platelets
Fatty acid Controls NSCLC patients Statistical significance Controls NSCLC patients Statistical significance
14:0 2.18 ±1.49 1.27 ±0.79 P<0.001 2.77 ±1.37 2.36 ±1.34 N.S.
15:0 0.42 ±0.39 0.38 ±0.36 N.S. 0.86 ±0.46 0.64 ±0.35 N.S.
16:0 21.46 ±9.82 14.33 ±5.25 P<0.001 16.41 ±5.39 15.07 ±7.46 N.S.
16:1 1.44 ±0.92 1.19 ±0.69 N.S. 2.42 ±0.85 2.53 ±1.25 N.S.
17:0 0.68 ±0.58 0.77 ±0.72 N.S. 0.89 ±0.29 0.72 ±0.36 N.S.
17:1 0.25 ±0.26 0.33 ±0.45 N.S. 0.57 ±0.23 0.63 ±0.29 N.S.
18:0 42.11 ±11.53 47.65 ±11.23 P<0.037 25.58 ±9.50 24.56 ±8.46 N.S.
18:1n9 24.32 ±10.01 17.80 ±5.48 P<0.003 20.37 ±6.23 22.03 ±6.24 N.S.
18:2n6 6.65 ±3.71 6.42 ±4.03 N.S. 15.66 ±10.57 17.55 ±8.93 N.S.
20:0 0.18 ±0.23 0.48 ±0.57 P<0.050 1.26 ±0.74 0.99 ±0.54 N.S.
20:1n9 0.71 ±0.72 0.60 ±0.71 N.S. 1.82 ±0.98 1.62 ±1.12 N.S.
20:2n6 0.06 ±0.23 0.30 ±0.61 N.S. N.D. N.D. N.S.
20:4n6 7.07 ±11.41 4.72 ±2.82 N.S. 5.07 ±4.95 5.22 ±4.40 N.S.
20:5n3 0.14 ±0.34 0.30 ±0.48 N.S. 0.61 ±0.46 0.51 ±0.33 N.S.
21:0 0.54 ±0.94 1.08 ±0.75 P<0.001 1.80 ±0.77 0.66 ±0.27 P<0.001
22:0 0.17 ±0.37 0.62 ±0.79 P<0.003 1.12 ±0.95 0.97 ±0.54 N.S.
22:1n9 0.16 ±0.43 0.39 ±0.53 P<0.012 1.59 ±0.58 1.00 ±0.46 P<0.041
22:4n6 0.70 ±1.12 0.60 ±0.77 N.S. N.D. N.D. N.S.
22:5n3 +24:0 1.51 ±1.96 0.58 ±0.85 P<0.050 N.D. N.D. N.S.
22:6n3 +24:1n9 1.18 ±1.56 1.09 ±1.07 N.S. N.D. N.D. N.S.
SFA 55.34 ±19.83 63.56 ±13.03 N.S. 51.06 ±7.35 45.04 ±8.74 N.S.
MUFA 25.48 ±11.12 18.74 ±7.41 P<0.005 26.65 ±6.75 25.19 ±7.02 N.S.
PUFA 16.17 ±12.87 13.63 ±5.65 N.S. 22.13 ±10.05 26.36 ±10.17 N.S.
TOTAL UFA 41.65 ±19.63 32.38 ±7.64 P<0.041 47.75 ±8.39 51.56 ±5.32 N.S.
SFA/UFA 1.73 ±1.06 2.12 ±1.05 N.S. 1.07 ±0.29 0.89 ±0.24 N.S.
UI 79.93 ±52.12 61.47 ±17.57 N.S. 83.64 ±19.63 95.19 ±18.73 N.S.
Data are given in percentages of total fatty acid contents and are means ±S.D. with GLC determinations performed in triplicate. N.D., not detected;
N.S., not significant; MUFA, monounsaturated fatty acids; PUFA, polyunsaturated fatty acids; SFA, saturated fatty acids; UFA, unsaturated fatty
acids; UI, unsaturation index, calculated as the sum of the percentage by weight of each fatty acid times the number of olefinic bonds.
PS +PI, 16:0 (89 and 74%, respectively). Platelets: PC, 20:5n3
(100 and 80%, respectively); PE, 20:4n6 and 22:5n3 +24:0 (77
and 100%, and 100 and 80%, respectively); PS +PI, 21:0 (91
and 80%); SM, 20:1n5 and 22:1n9 (100 and 77%, and 100 and
80%, respectively).
DISCUSSION
It is acknowledged that different biochemical parame-
ters and serum tumor markers have several potential appli-
cations in clinical oncology, such as screening, diagnosis,
prognosis, monitoring responses to therapy, or assessment of
the spontaneous course of the disease. For many years, lung
serum tumor markers, including a heterogeneous group of sub-
stances, have been used for a wide range of clinical applica-
tions (27,28), although high concentrations of these markers are
mainly found at advanced stages of the disease (29). Among
these serum markers, soluble degradation products of cytoker-
atins, such as TPS and Cyfra 21-1, are measurable in the periph-
eral blood of patients and are able to mark the existence of cancer
(27). Furthermore, the high sensitivity of sialic acid as a tumor
marker has been reported in a variety of cancerous conditions
(3). The relevance of sialic acid to the tumor cell is apparent from
the increased sialylation and sialyltransferase activity observed
in many cancer cells (3). Also, metabolomics, and specifically
plasma fatty acid profiling, has increasingly been used in many
fields, including disease diagnosis (10).
Since we have previously described that advanced NSCLC
is associated with changes in the fatty acids of total lipids from
peripheral erythrocytes and platelets (15), here we analyzed the
fatty acid profiles of phospholipid species in erythrocytes and
platelets from this type of patient in order to detect potential
fatty acids that could eventually be used as biomarkers. For
comparison, we also examined the previously described markers
for lung carcinoma: sialic acid, TPS and Cyfra 21-1.
Cyfra 21-1, TPA and TPS are closely related markers. TPA
and Cyfra 21-1 are known to provide fairly similar informa-
tion owing to the presence of a common antigenic determi-
nant (cytokeratin-19) (27). TPS was proposed as a step for-
ward in comparison with TPA because it is highly correlated
with the proliferation rate of cancer cells (27). Several stud-
ies of patients with lung lesions (revised in 27) have found
that both TPS and Cyfra 21-1 are elevated in all cell types of
lung cancer, are more sensitive in advanced stages of disease,
are useful in monitoring the course of disease, and are prog-
nostically meaningful. It has been reported that TPS sensitivity
rates range between 13 and 54%, with elevated values of TPS
412 J. de Castro et al.
Downloaded By: [Sánchez-Yagüe, Jesús] At: 11:16 29 April 2008
Tab l e 6. Changes in the fatty acid composition of sphingomyelin from erythrocytes and platelets in control subjects and NSCLC patients.
Erythrocytes Platelets
Fatty acid Controls NSCLC patients Statistical significance Controls NSCLC patients Statistical significance
14:0 1.41 ±0.55 1.10 ±0.67 P<0.016 2.12 ±0.59 2.24 ±0.92 N.S.
15:0 0.35 ±0.19 0.35 ±0.30 N.S. 0.77 ±0.26 0.55 ±0.30 N.S.
16:0 30.98 ±3.98 32.54 ±6.43 N.S. 17.37 ±3.29 29.71 ±12.40 P<0.001
16:1n7 0.82 ±0.70 0.60 ±0.55 N.S. 3.04 ±1.19 2.35 ±1.03 N.S.
17:0 0.71 ±0.38 0.83 ±0.65 N.S. 1.02 ±0.59 1.02 ±0.33 N.S.
17:1 ND ND 0.67 ±0.28 0.39 ±0.02 P<0.024
18:0 14.81 ±6.54 15.64 ±7.65 N.S. 14.31 ±3.97 15.50 ±3.97 N.S.
18:1n9 8.53 ±6.28 5.35 ±5.33 P<0.024 17.11 ±4.33 16.97 ±11.07 N.S.
18:2n6 3.71 ±2.54 4.36 ±4.85 N.S. 19.67 ±7.81 11.71 ±3.66 P<0.015
20:0 1.49 ±0.38 1.30 ±0.51 N.S. 1.95 ±0.77 2.37 ±2.47 N.S.
20:1n9 0.22 ±0.32 0.32 ±0.47 N.S. 3.35 ±1.19 1.01 ±0.62 P<0.001
20:2n6 0.06 ±0.21 0.01 ±0.08 N.S. N.D. N.D. N.S.
21:0 0.17 ±0.34 0.24 ±0.46 N.S. 2.05 ±0.68 1.70 ±1.62 N.S.
22:0 7.13 ±2.27 5.85 ±2.21 P<0.035 3.54 ±1.36 4.95 ±3.35 N.S.
22:1n9 0.28 ±0.49 0.17 ±0.46 N.S. 2.40 ±0.95 0.92 ±0.38 P<0.008
22:2n6 0.43 ±0.76 0.53 ±0.75 N.S. N.D. N.D. N.S.
23:0 1.35 ±0.37 1.49 ±1.01 N.S. N.D. N.D. N.S.
24:0 17.95 ±4.37 16.81 ±6.03 N.S. 3.42 ±2.07 3.93 ±4.19 N.S.
24:1n9 15.99 ±6.07 15.34 ±6.51 N.S. 5.46 ±4.48 5.31 ±3.17 N.S.
SFA 71.14 ±8.03 71.80 ±13.60 N.S. 45.57 ±4.17 59.25 ±14.26 P<0.002
MUFA 23.88 ±6.85 22.27 ±8.55 N.S. 29.91 ±5.66 26.20 ±10.27 N.S.
PUFA 4.85 ±2.75 5.02 ±4.87 N.S. 22.63 ±9.11 12.56 ±3.35 P<0.009
TOTAL UFA 28.72 ±7.95 25.29 ±9.84 N.S. 51.12 ±7.80 36.45 ±14.94 P<0.009
SFA/UFA 2.75 ±1.09 3.56 ±2.53 N.S. 0.91 ±0.24 1.19 ±0.47 N.S.
UI 34.20 ±10.28 30.73 ±13.25 N.S. 76.07 ±20.06 50.47 ±17.70 P<0.009
Data are given in percentages of total fatty acid contents and are means ±S.D. with GLC determinations performed in triplicate. N.D., not detected;
N.S., not significant; MUFA, monounsaturated fatty acids; PUFA, polyunsaturated fatty acids; SFA, saturated fatty acids; UFA, unsaturated fatty
acids; UI, unsaturation index, calculated as the sum of the percentage by weight of each fatty acid times the number of olefinic bonds.
occurring more often in the most advanced stages (27). More-
over, Cyfra 21-1 seems to show higher accuracy in compari-
son with TPS in both small cell lung cancer and NSCLC (30).
In the present work, higher levels of sensitivity were observed
for TPS and at the best cutt-off point the sensitivity of both
TPS and Cyfra 21-1 were quite similar (70 and 68%, respec-
tively). Also, TPS showed higher areas under the ROC curves
than Cyfra 21-1 (0.76 and 0.68, respectively). Nevertheless, as
previously described [30] at commonly reported cut-off values
(140 U/L and 3.6 ng/mL for TPS and Cyfra 21-1, respectively),
Cyfra 21-1 showed higher sensitivity, PPV and accuracy than
TPS.
It is known that both TSA and BSA are increased in a variety
of tumors, including lung cancer, not only in plasma, but also in
BAL fluid and pleural effusions (31). Importantly, TSA levels
are not altered by the smoking habit (6), one of the main causes
of lung cancer. Although most work has concluded that TSA and
BSA are highly sensitive markers of lung cancer, their specificity
seems to be low (3). Nevertheless, the usefulness of sialic acid
seems clear in the monitoring of responses to treatment in pa-
tients and in the detection of recurrence in a variety of tumors,
including lung carcinoma (3). At the designated cut-off levels
used in this work (chosen on the basis of the highest accuracy),
the sensitivity of TSA or BSA (86 and 84%, respectively) was
similar to that reported previously (6, 32). Also, at these cut-off
levels TSA/TP and BSA/TP showed lower specificity than TSA,
FSA or BSA. Nevertheless, it has previously been reported that
TSA/TP is a more tumor-specific marker than TSA (33, 34).
This difference can be explained because in the present work
TP in serum remained unchanged, and TSA/TP seemed to be
a better tumor marker than TSA in patients with decreases in
serum TP levels, perhaps as a result of the particular tumor type,
metastatic phenomena, or metabolic alterations.
It has been described that in a series of human pathologies
sharing redox alterations as a hallmark, significant modifica-
tions of erythrocyte integrity and function occurs (35–37). Pre-
viously, we also described that in advanced NSCLC not only
total fatty acids, but also some membrane proteins, including
band 3 and glycophorins, change significantly (38). In platelets
from patients with advanced NSCLC, increased protein oxida-
tion has also been reported (15). Additionally, several studies
have described different changes in the fatty acid composition
of erythrocyte total lipids in patients with different types of can-
cer (13), including lung cancer (39), although we are not aware
of any studies describing erythrocyte fatty acids from individual
phospholipids. With respect to platelets, we are only aware of
one report addressing fatty acid profiles in isolated PC, PE and
PS +PI phospholipids (40). Although in that report, patients
suffered exclusively from squamous cell lung carcinoma and
staging was different from the one used here, we have coincident
Fatty Acids as Markers of Lung Cancer 413
Downloaded By: [Sánchez-Yagüe, Jesús] At: 11:16 29 April 2008
Tab l e 7. Changes in the fatty acid composition of phosphatidic acid
from erythrocytes in control subjects and NSCLC patients
Lung cancer Statistical
Fatty acid Controls patients significance
14:0 2.59 ±1.06 2.35 ±1.27 N.S.
15:0 0.63 ±0.38 0.47 ±0.43 N.S.
16:0 24.19 ±5.39 27.08 ±8.30 N.S.
16:1n7 3.03 ±1.15 2.70 ±0.94 N.S.
17:0 0.78 ±0.37 0.72 ±0.50 N.S.
18:0 20.12 ±6.72 21.45 ±6.08 P<0.049
18:1n9 24.14 ±4.20 27.72 ±6.02 P<0.002
18:2n6 20.52 ±8.46 13.33 ±8.41 P<0.0134
20:0 1.01 ±0.94 0.22 ±0.29 P<0.003
20:1n9 1.00 ±1.15 0.83 ±0.92 N.S.
20:2n6 0.04 ±0.17 0.11 ±0.37 N.S.
20:4n6 1.60 ±1.09 2.36 ±2.47 N.S.
20:5n3 0.97 ±0.91 0.15 ±0.26 P<0.001
21:0 1.35 ±0.67 1.12 ±0.95 N.S.
22:0 1.34 ±1.24 0.36 ±0.58 P<0.001
22:1n9 1.10 ±0.74 0.36 ±0.58 P<0.001
22:4n6 0.17 ±0.40 0.19 ±0.55 N.S.
22:5n3 +24:0 0.82 ±1.28 0.39 ±0.62 N.S.
22:6n3 +24:1n9 0.39 ±0.76 0.32 ±0.62 N.S.
SFA 46.91 ±9.81 50.84 ±11.17 N.S.
MUFA 27.14 ±7.51 31.61 ±6.58 N.S.
PUFA 25.85 ±8.09 17.22 ±10.31 P<0.006
TOTAL UFA 52.99 ±9.71 48.83 ±10.77 N.S.
SFA/UFA 0.95 ±0.38 1.15 ±0.54 N.S.
UI 89.84 ±16.61 73.70 ±23.84 P<0.021
Data are given in percentages of total fatty acid contents and are
means ±S.D. with GLC determinations performed in triplicate. N.S.,
not significant; MUFA, monounsaturated fatty acids; PUFA,
polyunsaturated fatty acids; SFA, saturated fatty acids; UFA,
unsaturated fatty acids; UI, unsaturation index, calculated as the sum
of the percentage by weight of each fatty acid times the number of
olefinic bonds.
data regarding decreases in the main phosphoglyceride fractions
of n3 fatty acids. The main differences are that our decreases in
20:5n3 and 22:5n3 seem to be phospholipid-specific (PC and PE,
respectively), although those authors did find such decreases in
both phospholipids. In our case, no changes in 22:6n3 fatty acids
were detected. These data could be physiologically relevant be-
cause it is known that platelets with low levels of n3 PUFA
produce more thromboxane B2 when stimulated with thrombin
(41), and we (15) and others (40) have reported platelet activa-
tion in lung cancer, as shown by increased β-thromboglobulin
concentration in plasma (15, 40) and by increased thromboxane
B2 formation by platelets (40). The PUFA depletion described
here could be due to different phenomena. First, the contribution
of phospholipase A2does not seem to be important because we
have already observed that the erythrocyte or platelet phospho-
lipid profile remains unchanged in advanced NSCLC, with no
significant increases in lysophospholipid percentages (15). The
second phenomenon involves fatty acid metabolic alterations in
tissues from NSCLC patients. In cancer, changes in Delta 9 and
Delta 6 desaturases have been suggested (13), and in our case, a
decrease in such enzymes could at least partially explain the de-
creased levels of PUFA (42). A final possibility would involve
lipid peroxidation, a phenomenon well known to be linked to
both carcinogenesis and tumor behavior (43) that finally pro-
duces fluorescent chromolipids. In this regard, we have previ-
ously described that in advanced NSCLC, at least in platelets, the
decrease in PUFA from total lipids is correlated with an increase
in lipid fluorescence (15). Thus, a comparison of the changes in
the fatty acids of total lipids from erythrocytes and platelets
(15), together with the data reported here, lead us to propose,
for advanced NSCLC, the phospholipid(s) that the changes in
some of the main fatty acids from total lipids are due to. Thus,
in erythrocytes the increase in 18:0 from total lipids is due to PE
and PS +PI (mainly to PE, considering that it represents a 31%
of total phospholipids, while PS +PI represents a 9%, [15]).
The decrease in 18:2n6 is due to PC. The decrease in 20:4n6 is
due to PC and PE (mainly to PE considering the percentage of
20:4n6 in this phospholipid). The decrease in 22:6n3 +24:1n9
is due to PC and PE. The decrease in PUFA and IU, and the
increase in MUFA is also due to PC and PE. In platelets, the
increase in 16:0 and SFA from total lipids is due both to PC
and SM.
In spite of all previous evidence concerning changes in the
lipid composition of erythrocytes and platelets in cancer, no
efforts have been made to analyze whether the fatty acids from
those blood cells could be used as potential biomarkers in cancer
in general, and lung cancer in particular. Accordingly, analysis
of the operating characteristics for the well established serum
markers sialic acid and cytokeratins led us to compare them
with those obtained for some of the erythrocyte- and platelet-
specific phospholipid fatty acids than change in NSCLC pa-
tients. We considered it more appropriate to look mainly for
fatty acids that decrease rather than increase because, as indi-
cated above, changes in desaturases and lipid peroxidation pro-
cesses occurring in cancer will always elicit PUFA depletion. In
this regard, we detected some potential useful fatty acid mark-
ers. Some could be used because their changes are cell-specific,
and their diagnostic accuracy is high: Erythrocytes: PC, 18:2n6
and 20:4n6; PE, 22:4n6 and 22:6n3 +24:1n9. Platelets: PC,
22:0; PE, 22:5n3 +24:0. Another one is 20:5n3 of PC, a phys-
iologically relevant species, although not cell-specific. At the
cut-off value to obtain maximum accuracy, the best cell-specific
biomarkers are located in platelets: PS +PI, 21:0; SM:20:1n9
and 22:1n9. In erythrocytes, the best candidate is 16:0 in PS +PI.
Now that potential individual fatty acid biomarkers have been
detected we can target our research more precisely. In this re-
gard, it will be necessary to increase the size of the sample and to
include patients with benign lung pathologies, mainly of inflam-
matory nature, to ascertain the best applications of erythrocyte
or platelet fatty acid analysis in relation to screening, diagnosis
(either of tumors, cell type, or stage of disease), prognosis, mon-
itoring responses to anti-cancer therapies and follow-up of the
disease. Efforts are now under way to elucidate some of these
features and draw the pertinent conclusions.
It will also be interesting to compare the potential fatty
acid biomarkers found in this work using ROC operating
characteristics with those found using other, more complex
414 J. de Castro et al.
Downloaded By: [Sánchez-Yagüe, Jesús] At: 11:16 29 April 2008
Tab l e 8. Areas under the ROC curves for erythrocyte- and platelet-specific phospholipid fatty acids, and serum sialic
acids and cytokeratins
CI 95%
Area Lower limit Upper limit
Blood cell type Erythrocyte Platelet Erythrocyte Platelet Erythrocyte Platelet
PC
16:0 0.75 0.54 0.96
18:1n9 0.68 0.81 0.54 0.62 0.81 1.00
18:2n6 0.81 0.69 0.91
20:0 0.76 0.55 0.97
20:1 0.78 0.58 0.98
20:4n6 0.85 0.76 0.94
20:5n3 0.78 0.93 0.65 0.77 0.91 1.084
22:0 0.96 0.86 1.059
22:2 0.68 0.55 0.82
22:4n6 0.68 0.54 0.81
22:6n3 +24:1n9 0.73 0.61 0.85
PE
16:0 0.70 0.56 0.83
18:0 0.67 0.75 0.53 0.56 0.80 0.95
18:1n9 0.69 0.56 0.83
20:0 0.88 0.72 1.034
20:4n6 0.79 0.80 0.67 0.59 0.91 1.007
22:1n9 0.92 0.77 1.065
22:4n6 0.84 0.74 0.94
22:5n3 +24:0 0.93 0.78 1.088
22:6n3 +24:1n9 0.82 0.72 0.93
PS +PI
16:0 0.78 0.64 0.92
18:0 0.66 0.51 0.81
18:1n9 0.72 0.58 0.86
21:0 0.97 0.91 1.032
22:1n9 0.77 0.56 0.98
22:5n3 +24:0 0.63 0.50 0.77
SM
14:0 0.68 0.55 0.81
16:0 0.91 0.80 1.025
18:1n9 0.67 0.53 0.80
18:2n6 0.82 0.63 1.001
20:1n9 0.96 0.88 1.048
22:0 0.66 0.53 0.79
22:1n9 0.97 0.88 1.061
PA
18:1n9 0.77 0.62 0.92
18:2n6 0.72 0.57 0.87
20:0 0.75 0.59 0.91
20:5n3 0.93 0.85 1.01
22:1n9 0.80 0.66 0.94
Serum markers
Sialic Acids
TSA 0.79 0.65 0.94
FSA 0.67 0.52 0.83
BSA 0.79 0.65 0.94
TSA/TP 0.81 0.68 0.94
BSA/TP 0.76 0.61 0.92
Cytokeratins
TPS 0.76 0.63 0.89
Cyfra 21-1 0.68 0.54 0.83
Fatty Acids as Markers of Lung Cancer 415
Downloaded By: [Sánchez-Yagüe, Jesús] At: 11:16 29 April 2008
Tab l e 9. Diagnostic characteristics of erythrocyte- and platelet-specific phospholipid fatty acids, and serum
sialic acids and cytokeratins
Cut-off point Sensitivity Specificity PPV NPV Accuracy
Blood cell type Ery Pla Ery Pla Ery Pla Ery Pla Ery Pla Ery Pla
Phospholipid fatty acid
PC
16:0 25 77 75 77 75 76
18:1n9 21 24 69 77 57 77 68 77 59 77 63 77
18:2n6 19 11 81 50 68 67 78 64 71 53 75 57
20:0 0.7 70 58 58 70 63
20:1 1.15 75 67 60 80 70
20:4n6 2.47965757173
20:5n3 0.18 0.4 56 100 90 80 83 86 69 100 74 92
22:0 0.7 83 88 83 88 86
22:2
22:4n6
22:6n3 +24:1n9 0.81 65 71 71 65 68
PE
16:0 25 70 63 76 56 67
18:0 14 19 64 79 54 55 69 69 48 67 60 68
18:1n9 28 65 63 76 50 64
20:0 1.3 80 78 80 78 79
20:4n6 8 11 71 77 77 100 85 100 59 73 73 86
22:1 0.7 86 86 86 86 86
22:4n6 3 75 65 77 62 71
22:5n3 +24:0 1.75 100 80 86 100 91
22:6n3 +24:1n9 2.17171806071
PS+PI
16.0 20 89 74 85 81 84
18:0 38 79 55 75 60 70
18:1n9 21 78 69 78 69 74
21:0 0.71 8291 67 80 85 83 63 89 77 86
22:1n9 0.7 1.3 42 70 60 70 80 70 21 70 46 70
22:5n3 +24:0 0.624100 100 58 63
SM
14:0 1.26352655058
16:0 21 82 77 75 83 79
18:1n9 7 58 64 70 50 60
18:2n6 14 89 75 73 90 81
20:1n9 2 100 77 67 100 84
22:0 6.56494965261
22:1n9 1.5 100 80 80 100 91
PA
18:1n9 24 81 67 78 71 75
18:2n6 16.57163686767
20:5n3 0.65780578073
22:1n9 0.75 50 80 57 75 69
Serum markers
Sialic acids
TSA 2.25 86 78 90 70 84
FSA 0.027 95 61 85 85 70
BSA 2.25 84 83 92 68 84
TSA/TP 0.025 88 44 79 62 74
BSA/TP 0.025 86 44 79 57 74
Cytokeratins
TPS 37 70 71 87 48 70
Cyfra 21-1 1 68 57 81 40 65
The best cut-off points (given as percentages of total fatty acid content) were chosen on the basis of the highest
accuracy. Ery, erythrocytes; Pla, platelets; PPV, positive predictive value; NPV, Negative predictive value.
416 J. de Castro et al.
Downloaded By: [Sánchez-Yagüe, Jesús] At: 11:16 29 April 2008
statistical or mathematical approaches, such as supervised par-
tial least squares-linear discrimination analysis (PLS-LDA)
method. This multivariate approach has recently been success-
fully employed to detect plasma fatty acid biomarkers in type-2
diabetes mellitus patients after analysis of the plasma fatty acid
profiles using GC/MS (14).
ACKNOWLEDGMENT
The authors wish to thank N. Skinner for his assistance in the
preparation of the manuscript.
REFERENCES
1. Yogeeswaran, G. Cell surface glycolipids and glycoproteins in ma-
lignant transformation. Adv Cancer Res 1983,38, 289–350.
2. Schauer, R. Sialic acid as antigenic determinants of complex car-
bohydrates. Adv Exp Med Biol 1988,228, 47–72.
3. Narayanan, S. Sialic acid as a tumor marker. Ann Clin Lab Sci
1994,24 (4), 376–384.
4. Patel, P.S.; Baxi, B.R.; Balar, D.B. Significance of serum sialogly-
coproteins in patients with lung cancer. Neoplasma 1989,36 (1),
53–59.
5. Stringou, E.; Chondros, K.; Kouvaris, J.; Kakari, S.; Papavassil-
iou, K. Serum sialic acid (TSA/LSA) and carcinoembryonic anti-
gen (CEA) levels in cancer patients undergoing radiotherapy. An-
ticancer Res 1992,12 (1), 251–256.
6. Patel, P.S.; Raval, G.N.; Rawal, R.M.; Patel, G.H.; Balar, D.B.; Shah
P.M.; Patel, D.D. Comparison between serum levels of carcinoem-
bryonic antigen, sialic acid and phosphohexose isomerase in lung
cancer. Neoplasma 1995,42 (15), 271–274.
7. Moll, R.; Franke, W.W.; Schiller, D.L.; Geiger, B.; Krepler, R. The
catalog of human cytokeratins: patterns of expression in nor-
mal ephitelia, tumors and cultured cells. Cell 1982,31 (1), 11–
24.
8. Buccheri, G.; Ferrigno, D. Cytokeratin-derived markers of lung can-
cer. Expert Rev Mol Diagn 2001,1(3) , 315–322.
9. Chen, M.; Hofestadt, R. A medical bioinformatics approach
for metabolics disorders. Biomedical data prediction, modelling,
and systematic analysis. J Biomed Inform 2006,39 (2), 147–
159.
10. Mosconi, C.; Agradi, E.; Gambetta, A.;. Bozzetti, F.; Galli, C.
Decrease of polyunsaturated fatty acids and elevation of the
oleic/stearic acid ratio in plasma and red blood cell lipids of mal-
nourished cancer patients. JPEN J Parenter Enteral Nutr 1989,13
(5), 501–504.
11. Newcomer, L.M.; King, I.B.; Wicklund, K.G.; Stanford, J.L. The as-
sociation of fatty acids with prostate cancer risk. Prostate 2001,47
(4), 262–268.
12. Pala, V.; Krogh, V.; Muti, P.; Chajes, V.; Riboli, E.; Micheli, A.; Saa-
datian, M.; Sieri, S.; Berrino, F. Erythrocyte membrane fatty acids
and subsequent breast cancer: a prospective Italian study. J Natl
Cancer Inst 2001,93 (14), 1088–1095.
13. Mikirova, N.; Riordan, H.D.; Jackson, J.A.; Wong, K.; Miranda-
Massari, J.R.; Gonzalez, M.J. Erythrocyte membrane fatty acid
composition in cancer patients. P R Health Sci J 2004,23 (2),
107–113.
14. Yi, L.Z.; He, J.; Liang, Y.Z.; Yuan D.L.; Chau, F.T. Plasma fatty
acid metabolic profiling and biomarkers of type 2 diabetes mellitus
based on GC/MS and PLS-LDA. FEBS Lett 2006,580 (30), 6837–
6845.
15. de Castro, J.; Hern´andez-Hern ´andez, A.; Rodr´ıguez, M.C.; Llanillo,
M.; S´anchez-Yag¨ue, J. Comparison of changes in erythrocyte and
platelet fatty acid composition and protein oxidation in advanzed
non-small cell lung cancer. Cancer Invest 2006,24 (4), 339–345.
16. Zieba, M.; Suwalski, M.; Kwiatkowska, S.; Piasecka, G.;
Grzelewska-Rzymowska, I.; Stolarek, R.; Nowak, D. Comparison
of hydrogen peroxide generation and the content of lipid peroxida-
tion products in lung cancer tissue and pulmonary parenchyma.
Respir Med 2000,94 (8), 800–805.
17. Kinnula, V.L.; Paakko, P.; Soini, Y. Antioxidant enzymes and redox-
regulating thiol proteins in malignancies of human lung. FEBS Lett
2004,569 (1–3), 1–6.
18. Hernandez-Hernandez, A.; Llanillo, M.; Rodriguez, M.C.; Gomez,
F. ; Sanchez-Yague, J. Amphiphilic and hydrophilic nature of sheep
and human platelet phosphotyrosine phosphatase forms. Biochim.
Biophys Acta 1999,1419 (2), 195–206.
19. Hernandez-Hernandez, A.; Garabatos, M.N.; Rodr´ıguez, M.C.;
Vidal, M.L.; L ´opez-Revuelta, A.; S´anchez-Gallego, J.I.; Llanillo, M.;
Sanchez-Yag ¨ue, J. Structural characteristics of a lipid peroxida-
tion product, trans-2-nonenal, that favour inhibition of membrane-
associated phosphotyrosine phosphatase activity. Biochim Bio-
phys Acta 2005,1726 (3), 317–325.
20. Warren, L. The thiobarbituric acid assay of sialic acids. J Biol Chem
1959,234 (8), 1971–1975.
21. Aminoff, D. Methods for the quantitative estimation of N-
acetylneuraminic acid and their application to hydrolysates of sialo-
mucoids. Biochem J 1961,81 (Nov), 384–392.
22. Gornall, A.G.; Bardawill, C.J.; David, M.M. Determination of serum
proteins by means of Biuret reaction. J Biol Chem 1949,177 (2),
751–766.
23. L´opez-Revuelta, A.; S´anchez-Gallego, J.I.; Hern ´andez-
Hern ´andez, A.; S ´anchez-Yag ¨ue, J.; Llanillo, M. Membrane
cholesterol contents influence the protective effects of quercetin
and rutin in erythrocytes damaged by oxidative stress. Chem Biol
Interact 2006,161 (1), 79–91.
24. L´opez-Revuelta, A.; S´anchez-Gallego, J.I.; Hern ´andez-
Hern ´andez, A.; S ´anchez-Yag ¨ue, J.; Llanillo, M. Increase in
vulnerability to oxidative damage in cholesterol-modified erythro-
cytes exposed to t-BuOOH. Biochim Biophys Acta 2005,1734 (1),
74–85.
25. de Castro, J.; Hern´andez-Hern ´andez, A.; Rodr´ıguez, M.C.; Sar-
dina, J.L.; Llanillo, M.; S´anchez-Yag¨ue, J. Comparison of changes
in erythrocyte and platelet phospholipid and fatty acid composition
and protein oxidation in chronic obstructive pulmonary disease and
asthma. Platelets 2007,18(1), 43–51.
26. Kaplan K.L.; Owen, J. Plasma levels of beta-thromboglobulin and
platelet factor 4 as indices of platelet activation in vivo. Blood 1981,
57 (2), 199–202.
27. Buccheri, G.; Ferrigno, D. Lung tumor markers of cytokeratin origin:
an overview. Lung Cancer 2001,34 (Suppl. 2), S65–S69.
28. Fischer, J.R.; Lahm, H. Validation of molecular and immunological
factors with predictive importance in lung cancer. Lung Cancer
2004,45 (Suppl. 2), S151-S161.
29. Plabani, M.; Basso, D.; Navaglia, F.; De Paoli, M.; Tommasini, A.;
Cipriani, A. Clinical evaluation of seven tumor markers in lung can-
cer diagnosis: can any combination improbe the results?. Br J Can-
cer 1995,72 (1), 170–173.
30. Pujol, J.L; Grenier, J.; Parrat, E.; Lehmann, M.; Lafontaine, T.;
Quantin X.; Michel, F.B. Cytoqueratins as serum markers in lung
cancer: a comparison of Cyfra 21-1 and TPS. Am J Respir Crit
Care Med 1996,154 (3 Pt1), 725–733.
31. Alatas, F.; Alatas, ¨
O.; Metintas, M.; Colak, ¨
O.; Harmanci, E.; Demir,
S. Diagnostic value of CEA, CA 15-3, CA 19-9, Cygra 21-1, NSE
and TSA assay in pleural effusions. Lung Cancer 2001,31 (1),
9–16.
32. Kakari, S.; Stringou, E.; Toumbis, M.; Ferderigos, A.S.; Poulaki, I.;
Chondros, K.; Dema, A.; Kotsovoulou, V.; Pavlidis, N. Five tumor
markers in lung cancer:significance of total and “lipid”-bound sialic
acid. Anticancer Res 1991,11 (6), 2107–2110.
Fatty Acids as Markers of Lung Cancer 417
Downloaded By: [Sánchez-Yagüe, Jesús] At: 11:16 29 April 2008
33. Plucinsky, M.C.; Riley, W.M.; Prorok, J.J.; Alhadeff, J.A. Total and
lipid associated serum sialic acid levels in cancer patients with dif-
ferent primary sites and different degree of metastatic involvement.
Cancer 1986,58 (12), 2680–2685.
34. Feijoo-Carnero, C.; Rodr´ıguez-Berrocal, F.J.; P ´aez de la Cadena,
M.; Ayude, D.; de Carlos, A. Mart´ınez-Zorzano, V.S. Clinical signifi-
cance of preoperative serum sialic acid levels in colorectal cancer:
utility in the detection of patients at high risk of tumor recurrence.
Int J Biological Markers 2004,19 (1), 38–45.
35. Straface, E.; Masella, R.; Del Principe, D.; Franceschi, C.; Korkina,
L.G.; Zatterale, A.; Pagano, G.; Malorni, W. Spectrin changes oc-
cur in erythrocytes from patients with Fanconi’s anemia and their
parents. Biochem Biophys Res Commun 2000,273 (3), 899–
901.
36. Straface, E.; Matarrese, P.; Gambardella, L.; Forte, S.; Carlone, S.;
Libianchi, E.; Schmid, G.; Malorni, W. N-acetylcisteine counteracts
erythrocyte alterations occurring in chronic obstructive pulmonary
disease. Biochem Biophys Res Commun 2000,279 (2), 552–
556.
37. Straface, E.; Rivabene, R.; Masella, R.; Santulli, M.; Paganelli,
R.; Malorni, W. Structural changes of the erythrocyte as a
marker of non-insulin-dependent diabetes. Protective effects of
N-acetylcysteine. Biochem Biophys Res Commun 2002,290 (5),
1393–1398.
38. Hern ´andez-Hern ´andez, A.; Rodr´ıguez, M.C.; L´opez-Revuelta. A.;
S´anchez-Gallego, J.I.; Shnyrov, V.; Llanillo, M.; S´anchez-Yag¨ue,
J. Alterations in erythrocyte membrane protein composition in ad-
vanced non-small cell lung cancer. Blood Cell Mol Dis 2006,36
(3), 355–363.
39. Khyshiktuev, B.S.; Khyshiktueva, N.A.; Ivanov, V.N.; Darenskaia,
S.D.;Novikov, S.V. Fatty acid composition of blood plasma lipids
and erythrocytes in lung cancer patients. Vopr Med Khim 1994,40
(5), 48–50.
40. Prisco, D.; Paniccia, R.; Coppo, M.; Filippini, M.;
Francalanci, I.; Brunelli, T.; Comeglio, P.; Abbate, R. Platelet
activation and platelet lipid composition in pulmonary cancer.
Prostaglandins Leukot Essent Fatty Acids 1995,53 (1), 65–68.
41. Prisco, D.; Rogasi, P.G.; Paniccia, R.; Abbate, R.; Gensini, G.F.;
Pinto, S.; Vanni D.; Neri-Serneri, G.G. Altered membrane fatty acid
composition and increased thromboxane A2 generation in platelets
from patients with diabetes. Prostaglandins Leukot Essent Fatty
Acids 1989,35 (1), 15–23.
42. Nakamura, M.T.; Nara, T.Y. Structure, function, and dietary regu-
lation of delta6, delta5, and delta9 desaturases. Annu Rev Nutr
2004,24, 345–376.
43. Kinnula, V.L.; Paakko, P.; Soini, Y. Antioxidant enzymes and redox-
regulating thiol proteins in malignancies of human lung. FEBS Lett
2004,569 (1–3), 1–6.
418 J. de Castro et al.
