Does the association of 18F-FDG uptake intensity and lesion topography reveal histological phenotype and tumor differentiation in esophageal cancer?

Abstract

In daily clinical practice, the esophageal squamous cell cancer (ESCC) is considered to be more (18)F-FDG avid than adenocarcinoma (EAD). To date, the few studies concerning the existence of a real metabolic difference based on esophageal cancer (EC) histology, show divergent and not definitive results. A retrospective analysis of (18)F-FDG PET/CT of 87 patients with ESCC and EAD was performed to investigate the role played by both histopathological subtype and tumor differentiation in the characterization of glucose metabolic profile of EC. Esophageal squamous cell cancer was well differentiated (WD) in 42 cases and poorly differentiated (PD) in 12 patients. Twenty-one of the 33 patients had WD EAD, while 12 had a PD EAD. The (18)F-FDG maximal standardized uptake value (SUV(max)) was determined for all lesions and used for inter and intra-group comparison. In ESCC, the SUV(max) ranged from 4 to 31 with a mean value of 16±6. In EAD, the SUV(max) ranged from 2 to 25 with a mean value of 10±6. A statistically significant difference (P<0.0001) was found between ESCC and EAD. According to histological classification and tumor differentiation, we obtained the following results: a) the SUV(max) values of WD ESCC and WD EAD were 17±5 (range: 7-31) and 7±3 (range: 2-12) respectively (P<0.00001), b) the SUV(max) values of PD ESCC and PD EAD were 11±4 (range: 4-19) and 17±6 (range: 7-25) respectively (P<0.05). Moreover, a statistically significant difference of SUV(max) values was found between WD and PD ESCC (P<0.005) as well as between WD and PD differentiated EAD (P<0.0001). In order to predict tumor histology (ESCC, EAD) from both SUV(max) and lesion location, a multivariate discriminant analysis was performed on the whole population with a resulting diagnostic accuracy equal to 82% (P<0.00001). In conclusion, we provide additional arguments about (18)F-FDG uptake difference between ESCC and EAD as well as between poorly and well-differentiated forms of both EC histological subtypes.

Full text

Original Article
ww w.n uc lme d. gr 23 9
Hellenic Jour nal of Nucle ar Medic ine
Se pte m ber -De c emb er 2 011
Abstract
In daily clinical practice, the esophageal squamous cell cancer (ESCC) is
considered to be more 18F-FDG avid than adenocarcinoma (EAD). To date,
the few studies concerning the existence of a real metab olic difference
base d on esophageal cancer (EC) histology, show divergent and not de-
finitive results. A retros pective a nalysis o f 18F-FDG PET/CT of 87 patients
with ESCC and EAD was performed to investigate the role played by both
histopathologic al subtype and tumor differentiation in the characteriza-
tion of glucose metabo lic profile of EC . Esophageal squamous cell cancer
was well differentiated (WD) in 42 cases and po orly differentiated (PD) in
12 patients . Twent y-one of the 33 patients had WD EAD, while 12 had a
PD EAD. The 18F-FDG maximal standardized uptake value (SUVma x) was de-
termined for all lesions and used for inter and intra-group comparison. I n
ESCC, the SUVmax ra nged from 4 to 31 with a mean value of 16±6. In EAD, the
SUVma x ranged from 2 to 25 with a mean value of 10±6. A statistically signif-
icant difference (P<0.0001) was found between ESCC and EAD. According
to histol ogical classification and tumor diffe rentiation, we obtained the
following result s: a) the SUVma x values of WD ESCC and WD EAD were 17±5
(range:7-31) and 7±3 (range:2-12) respectively (P<0.00001), b) the SUVm ax
values of PD ESCC and PD EAD were 11±4 (range:4 -19) and 17±6 (range:7-
25) respective ly (P<0.05). Moreover, a statistically significant difference
of SUVmax values was found between WD and PD ESCC (P<0.005) as well
as between WD and PD differentiated EAD (P<0.0001). In order to predict
tumor histology (ESCC, EAD) from both SUVma x and lesion location, a multi-
variate discriminant analysis was performed on the whole population with
a resulting diagnostic accuracy equal to 82% (P<0.0 0001). In concl usion, we
provide additional arguments about 18F-FDG uptake difference between
ESCC and EAD as well as bet ween poorly and well-differentiated forms of
both EC histological subtypes.
Introduction
Ruorine-18-uorodeoxyglucose (18F-FDG) positron emission
tomography (PET) is nowadays widely applied in the man-
agement of esophageal cancer (EC). Esophageal cancer in-
cludes two major histological subtypes: the squamous cell cancer
(ESCC) and the adenocarcinoma (EAD) [1-2]. Tumor 18F-FDG uptake
intensity is directly related to both the expression of cellular mem-
brane glucose transporter-1 protein (Glut-1) and the cellular gly-
colysis, so the 18F-FDG maximal standardized uptake value (SUVm ax)
could allow the estimation of the tumor glucose metabolism rate.
The relationship between SUVma x and both tumor histological type
and dierentiation has been assessed for various malignancies such
as lung and cervical cancer [3-4], but the literature oers limited
data concerning the EC [5].
In the present study, we have thus investigated the role played by
both histopathologic subtype and tumor dierentiation in the glu-
cose metabolic prole of EC.
Does the association of 18F-FDG uptake intensity and
lesion topography reveal histological phenotype and tumor
differentiation in esophageal cancer?
Alessio Imp eriale
1,2
MD, Sébastian Ci marelli
3
MD,
Cécile Brig and
4
MD, Guillaume Faure
5
MD,
Gilles Kar cher
6
MD, Serge Rohr
4
MD,
David Atlani
7
MD, Pierre Oliv ier
6
MD
1. Service de Biophysiq ue et de Médecine Nucléaire ,
Hôpital de Haute pierre, Hôpitaux Universitai res de
Strasbourg, Strasbourg, France
2. LINC, UMR 7237, UDS / CNRS, Strasbourg, France
3. Service de Mé decine Nucléaire, CLRCC Leon Berard,
Lyon, France
4. Service de Ch irurgie Générale et Digestive , Hôpital de
Hautepierre , Hôpitaux Universitaires de Strasbourg,
Strasbourg, France
5. Service de Rad iothérapie, Clinique Claud e Bernard,
Metz, France
6. Service de B iophysique et de Médecine Nuc léaire,
Hôpital de Brabo is, Hôpitaux Universitaires de Na ncy,
Vandoeuvre Lès Nanc y, France
7. Service de Radiothérapie , Hôpital Civil, Colmar, France
***
Keywords: Esophageal cancer
- FDG PET
- Esophageal adenocarcinoma
- Esophageal squamous cell carcinoma
- Tumor dierentiation
Corre spondence a ddress:
Alessio Imperiale, MD
Service de Biophysique et de Médecine Nucléaire,
Hôpital de Hautepierre, Hôpitaux Universitaires de
Strasbourg 1, Avenue Molière, 67098 Strasbourg
Cedex, France, Tel : +33388127552, Fax : +33388128121
Email : alessio.imperiale@chru-strasbourg.fr
Received:
5 September 2011
Accepted revised:
17 October 2011
Hell J Nucl Med 2011; 14(3): 239-242 Published on line: 10 November 2011

Original Article
Hellenic Jour nal of Nucle ar Medic ine
Se pte m ber -De c emb er 2 011 ww w.n uc lme d. gr
240
Original Article
Subjects and methods
We have performed a retrospective analysis of patients
addressed to the Nuclear Medicine Department of both
Strasbourg and Nancy University Hospitals for EC evaluation
by PET/CT before any treatment.
Eligible patients were identied by 18F-FDG PET/CT databas-
es according to the following inclusion criteria: a) either ESCC
or EAD proved by biopsy, b) tumor stage from II to IV proved by
either endoscopic ultrasound (EUS) or pathologic criteria ac-
cording to the TNM system of the American Joint Committee
on Cancer, c) availability of unequivocal pathological informa-
tion about tumor dierentiation: only well dierentiated (WD)
and poorly dierentiated (PD) tumors were selected.
We have left out of our study the patients with: a) histolog-
ical types of EC dierent from ESCC and EAD (i.e. leimyoma,
gastrointestinal stromal tumor, small cell carcinoma), b) tu-
mor stage 0 and I according to both clinical and pathological
criteria, c) either ESCC or EAD moderately dierentiated, and
d) history of esophageal surgery, radiotherapy or chemo-
therapy before PET examination.
Discovery (General Electric, Milwaukee, USA) and Biograph
Duo (Siemens, Knoxville, USA) PET/CT devices were used in
Strasbourg and Nancy, respectively. To obtain a serum glu-
cose level of less than 6.6mmol/L, the patient fasted for 6h be-
fore the intravenous injection of 5MBq/kg of 18F-FDG. Whole-
body PET/CT acquisitions started 60min after tracer injection,
including a head to mid thigh CT scan, followed by a 2-di-
mensional PET scan. Data from PET were reconstructed with
CT-based attenuation correction. SUVmax was determined as
follows:
SUVmax = [maximum pixel value in the tumor (kBq/mL)] / [in-
jected dose (kBq)/patient weight (g)].
The maximum pixel value in the tumor was obtained by
using circular regions of interest (ROI) covering the entire
tumor on trans-axial slices. These regions were selected by
two experienced nuclear medicine physicians (AI, PO) who
had been informed about the topography of the EC previ-
ously detected by the conventional diagnostic approach.
The results are expressed as mean±SD, median and range.
The Mann-Whitney U test was used for between-group com-
parisons. Comparison of proportions was done by the Fisher
exact test. To predict the tumor histology (ESCC, EAD) from
SUVmax and lesion location (proximal, medial and distal part of
esophagus), discriminant analysis was performed. To limit bias
in the number of patients correctly classied, the achieved
classication, after cross-validation, was submitted to a jack-
knife validation procedure. The Statistica package (STATSOFT;
www.statsoft.com) was used for the statistical data analysis. A
P<0.05 was considered statistically signicant.
Results
The clinical and 18F-FDG PET results are summarized in Table
1 and Table 2 respectively.
Table 1. Patient population clinical characteristics
No Sex Pt age Tumor location T T Di
(M / F) mean±SD (range) Up Med Low LC (II – IV) WD PD
ESCC 54 (39 / 15) 63±9 (49-82) 21 21 11 1 II 19 16 3
III 32 24 8
IV 3 2 1
EAD 33 (29 / 4) 62±11 (38-85) 1 2 21 9 II 14 11 3
III 19 10 9
IV - - -
ESCC: esophageal squamous cell cancer; EAD: esophageal adenocarcinoma; Up: upper esophagus; Med: medial esophagus; Low: lower
esophagus; LC: lower esophagus and cardia; T: tumor stage according to TNM classication [6]; T Di: tumoral dierentiation; WD: well
dierentiated; PD: poorly dierentiated.
Table 2. 18F-FDG PET results (SUVmax) obtained from the analysis of the whole population
WD EC PD EC
mean±SD median range mean±SD median range P WD vs. PD
ESCC 17±5 16 7-31 11±4 11 4-19 < 0.0001
EAD 7±3 6 2-12 17±6 17 7-25 < 0.005
P ESCC vs. EAD < 0.00001 < 0.05
ESCC: esophageal squamous cell cancer; EAD: esophageal adenocarcinoma; WD: well dierentiated; PD: poorly dierentiated; EC: es-
ophageal cancer.

ww w.n uc lme d. gr 24 1
Hellenic Jour nal of Nucle ar Medic ine
Se pte m ber -De c emb er 2 011
Discussion
The SUV index presents certain limitations that are mainly
due to important sources of variability in its determination
[6]. In spite of that, it is still considered as the reference index
whenever a quantitative evaluation is needed for diagnosis,
therapy evaluation or for the purpose of prognosis. A close
correlation between the 18F-FDG uptake intensity and both
the over expression of Glut-1 transmembrane transporters
and the up-regulation of intracellular hexokynase (HK) has
been previously assessed in EC [7-8]. In well-dierentiated
forms of lung and cervical cancer, the reduced Glut-1 and
HK-II expression are directly responsible for a low 18F-FDG
uptake [3-4]. It is then reasonable to suppose that cellular
dierentiation has a key role in the modulation of the 18F-
FDG uptake also in EC.
In daily clinical practice, ESCC is considered to be more
18F-FDG avid than EAD, the 18F-FDG uptake variability of EAD
appearing to be wider than that of the ESCC. But, to date,
the few studies concerning the existence of a real metabolic
dierence based on EC histology, show divergent and not
denitive results.
The rst study reporting a systematic investigation about
the inuence of histopathologic subtype and EC grading on
the 18F-FDG uptake was published by Mentzel et al (2003) [5].
These authors examined forty-six patients suering from
EC (28 ESCC and 18 EAD) by the pre-therapeutic 18F-FDG
PET/CT technique and dierent degrees of tumoral dier-
entiation. The SUVmax was used for 18F-FDG tumor intensity
evaluation. Both ESCC and EAD were characterized by an im-
portant intra-group variability in terms of SUVmax . Although
EAD showed a mean SUVmax value that was moderately less
than that of ESCC, the dierence of uptake intensity was
not statistically signicant. Likewise, there was a slight but
not relevant trend towards higher SUVmax in more dedier-
entiated cancer. Unfortunately, the size of all the examined
subgroups was limited, particularly those related to tumoral
dierentiation, so making the interpretation of the statistical
results both dicult and possibly not denitive. To explain
Among the eighty-seven patients selected for this study
(68 men and 19 women; age: 63±10; age range: 38-85), fty-
four (62%) and thirty-three (38%) patients were found to be
aected with ESCC and EAD respectively. Esophageal SCC
was well dierentiated in forty-two cases and poorly dif-
ferentiated in the remaining twelve patients. Twenty-one
of the 33 patients had well-dierentiated EAD, while 12 had
a poorly dierentiated EAD. Esophageal SCC was located in
the proximal, medial and distal part of the oesophagus in
21, 21 and 12 cases, respectively. On the other hand, EAD
was detected in the proximal, medial and distal part of the
oesophagus in 1, 2 and 30 patients. As for the T grade, 33,
51 and 3 esophageal lesions were graded as T2, T3 and T4
respectively.
All ESCC and EAD primary tumours were visualised on 18F-
FDG PET/CT images (Fig. 1).
Figure 1. Es ophageal well dierentiate d adenocarcinoma (A), poo rly dierenti-
ated adenocarcino ma (B), well dierentiated squam ous cell cancer (C) and poorly
dierentiated squamo us cell cancer (D). A representative image of each histologi-
cal type an d dierentiation is reporte d.
In the 54 examined ESCC, SUVmax ranged from 4 to 31 with
a mean value of 16±6. For all the EAD, SUVmax ranged from 2
to 25 with a mean value of 10±6. A statistically signicant dif-
ference (P<0.0001) was found between ESCC and EAD.
According to both histological classication and tumor
dierentiation (Fig. 2): a. The SUVmax values of WD ESCC
and EAD resulted as being equal to 17±5 (range:7-31) and
7±3 (range:2-12), respectively. A statistically signicant dif-
ference was found between ESCC and EAD (P<0.00001). b.
The SUVmax values of PD ESCC and EAD resulted as being
equal to 11±4 (range:4-19) and 17±6 (range:7-25), respec-
tively. A statistically signicant dierence was found be-
tween ESCC and EAD (P<0.05). c. A statistically signicant
dierence of SUVmax values was found between WD and
PD ESCC (P<0.005) as well as between WD and PD EAD
(P<0.0001).
In order to predict tumor histology (ESCC, EAD) from both
SUVmax and lesion location (LOC: proximal, medial and distal
part of the oesophagus), a multivariate discriminant analy-
sis was performed on the whole population (87 patients).
SUVmax together with LOC correctly identied 43 of 54 (80%)
ESCCs and 28 of 33 (85%) EAD with a global diagnostic ac-
curacy of 82%, a sensitivity and specicity of 72% and 90%,
respectively, an 85% positive predictive value, and an 80%
negative predictive value (P<0.00001).
Figure 2. Box pl ot of SUVmax values obt ained from the analysis of eight y-seven
esophageal cancer accord ing to both histopathology and tumor dierentiation.
Mean, standard error, standard deviation and outliers are graphically represented.
ESCC: es ophageal squamous cell cancer; EAD: esophageal adenocarcinoma; WD:
well dierentiate d; PD: poorly die rentiated.
Original Article
C. D.
A. B.
PD WD PD WD
24
18
12
6
SUVmax
EAD ESCC

Original Article
Hellenic Jour nal of Nucle ar Medic ine
Se pte m ber -De c emb er 2 011 ww w.n uc lme d. gr
24 2
the molecular events responsible for the increased 18F-FDG
uptake in EC, others [9] proposed a genetic model of es-
ophageal cancer. Interestingly, in their preliminary results,
the authors showed a signicantly increased 18F-FDG uptake
in ESCC compared to the EAD experimental model.
Our present study is focused on an accurately selected
population. Indeed, in order to minimize the PET partial vol-
ume eect on SUVmax estimation, tumor stage from II to IV
was only considered, so excluding both tumor stage 0 and
I. Moreover, in order to maximize the eect of dierentia-
tion on tumoral 18F-FDG uptake, we have not included any
moderately dierentiated ESCC or EAD. In spite of an im-
portant heterogeneity of SUVmax values, the non-parametric
statistical analysis applied to our data showed a signicant
dierence between the SUVmax of ESCC and that of EAD.
Furthermore, the good global diagnostic accuracy (82%)
when predicting tumor histology from both SUVmax and le-
sion location, underlined the existence of dierent glucose
metabolism between ESCC and EAD. The two histological EC
subtypes were even more evidently discriminated by adding
as a grouping factor the data regarding the tumor grading
to the statistical analysis. In the EAD case, the SUVmax was di-
rectly related to the tumoral dedierentiation. Conversely,
the more dierentiated the ESCC was, the more important
the 18F-FDG uptake intensity became, which agrees with the
observations of others [10]. These authors investigated the
expression of the neutral amino acid transporter ASCT1 and
its potential correlation with the Glut-1 glucose transporter
in forty-two resected EC. Interestingly, Glut-1 was expressed
more often in the well dierentiated ESCC than in the poorly
dierentiated one, representing a potential explication for
our ndings. Indeed, according to our results, the well dif-
ferentiated ESCC were characterized from higher values of
SUVmax than the poorly dierentiated one. The same authors
also showed that signicantly more EAD expressed ASCT1
than ESCC [10], suggesting dierent metabolic needs be-
tween these two tumor histological subtypes.
One of the major limitations of the present study, directly
related to its retrospective nature, is the lack of Glut-1 and
HK immunohistochemical quantication, which would have
allowed a better understanding of the relationship between
the tumoral pathophysiological mechanisms and SUVmax.
In conclusion, we provide additional arguments concern-
ing the 18F-FDG uptake dierence between ESCC and EAD
as well as between poorly and well-dierentiated forms of
both EC histological subtypes. Just for common opinion, our
results also suggest that poor dierentiation is not necessar-
ily matched with a high 18F-FDG uptake.
The authors decla re that they have no conicts of interest.
Bibliography
1. Enzinger PC, Mayer RJ. Esophageal cancer. N Engl J Med 2003;
349: 2241-52.
2. Luketich JD, Schauer PR, Meltzer CC et al. Role of positron emis-
sion tomography in staging esophageal cancer. Ann Thorac
Surg 1997; 64: 765-9.
3. Mamede M, Higashi T, Kitaichi M et al. 18F-FDG uptake and
PCNA, Glut-1, and Hexokinase-II expressions in cancers and
inammatory lesions of the lung. Neoplasia 2005; 7: 369-79.
4. Kidd EA, Spencer CR, Huettner PC et al. Cervical cancer histol-
ogy and tumor dierentiation aect 18F-uorodeoxyglucose
uptake. Cancer 2009; 1: 3548-54.
5. Menzel Ch, Döbert N, Rieker O et al. 18F-Deoxyglucose PET for
the staging of oesophageal cancer: inuence of histopatho-
logic subtype and tumour grading. Nuklearmedizin 2003; 42:
90-3.
6. Tylski P, Stute S, Grotus N et al. Comparative assessment of
methods for estimating tumor volume and standardized up-
take value in 18F-FDG PET. J Nucl Med 2010; 5: 268-76.
7. Tohma T, Okatsumi S, Makino H et al. Relationship between
glucose transporter, hexokinase and FDG-PET in esophageal
cancer. Hepatogastroenterology 2005; 52: 486-90.
8. Westerterp M, Sloof GW, Otto SH et al. 18FDG uptake in
oesophageal adenocarcinoma: linking biology and outcome.
J Cancer Res Clin Oncol 2008; 134: 227-36.
9. Mintz A, Kim SH, Alavi A et al. A genetic model of esophageal
cancer accurately portrays clinical FDG uptake. J Nucl Med
2006; 47(Suppl 1): 431P.
10. Younes M, Pathak M, Finnie D et al. Expression of the neutral
amino acids transporter ASCT1 in esophageal carcinomas.
Anticancer Res 2000; 20: 3775-80.