92 Neoplasma 59, 1, 2012

doi:10.4149/neo_2012_012

Characterization of adrenal metastatic cancer using FDG PET/CT

B. XU1, J. GAO2*, L. CUI2, H. WANG1, Z. GUAN1, S. YAO1, Z. SHEN1, J. TIAN1

Department of Nuclear Medicine, Chinese PLA General Hospital, Beijing, Peoples Republic of China; 2Department of Urology, Chinese PLA
1

General Hospital, Beijing, Peoples Republic of China

*Correspondence: [email protected]

Received August 8, 2011 / Accepted September 5, 2011

The adrenal gland is a common location for metastasis from a primary tumor in another organ. This study evaluated
the properties of adrenal lesions in cancer and non-cancer patients and investigated what variables may help predict
adrenal metastasis. This retrospective study used 18uorodeoxyglucose PET/CT on 371 patients with adrenal lesions (N
= 260 with a primary tumor and N = 111 with an unknown primary tumor). Parameters such as the presence of a tumor,
nodule, enlarged adrenal, maximum standardized uptake (SUVmax ratio) were evaluated. Univariate and multivariate
analysis were used to identify variables that may predict risk of adrenal metastasis. Subjects with adrenal metastasis versus
those without had a higher frequency of primary lung tumors (53.7% versus 28.6%, respectively; P 0.001) but a lower
frequency of gastrointestinal cancer (9.3% versus 20.4%, respectively; P = 0.014). The frequency of other abnormalities
including nodules and enlarged adrenals were similar between cancer and non-cancer subjects. A higher proportion of
subjects with adrenal metastasis regardless whether the primary tumor site in the lung, gastrointestinal track, or liver
had SUVmax ratio > 2.5 versus those with no adrenal metastasis. In this cohort of subjects, the greatest proportion of
subjects with adrenal metastasis was those with primary lung cancer. Univariate and multivariate analysis indicated
that age, SUVmax ratio, and the presence of metastasis in multiple organs were independent variables for having adrenal
metastasis. In this study, FDG PET/CT was useful in characterizing adrenal lesions including determining whether they
were benign or malignant. This technology allowed us to identify characteristics that may useful in predicting adrenal
metastasis and cancer severity.

Key words: uorodeoxyglucose, positron emission tomography, computed tomography, adrenal metastasis, cancer

The adrenal gland is a common site for metastasis in pa- procedure, however it is invasive and technically dicult
tients with cancer with the rate of metastasis being between [4]. Monitoring cortisol secretion is useful although not all
25% to 75% depending on the type and size of the primary malignant tumors result in increased hormone secretion and
tumor [1]. The most common malignant lesions that metasta- the increased secretion may result from other non-cancerous
size to the adrenal gland include malignant melanoma, breast, causes [2]. Imaging of the adrenal gland to diagnose a mass
lung, kidney, esophagus, pancreases, liver, stomach, and colon is typically accomplished using computed tomography (CT)
cancers [1,2]. However diagnosis of an adrenal lesion as ma- or magnetic resonance imaging (MRI). CT measures attenu-
lignant or benign can be problematic. For example, patients ation to dierentiate between benign and malignant lesions,
with adrenal metastasis are usually asymptomatic and even in but the use of attenuation is not always possible depending
patients with a history of cancer, 50% of adrenal tumors are upon the characteristic of the lesions [5,6]. The use of MRI can
not malignant [3]. Moreover, although most adrenal masses be problematic since the malignant and benign lesion signal
in patients are benign, about 2.5% are malignant [1]. Accurate intensity overlap [7].
characterization of these adrenal lesions in cancer patients is 18
F-Fluorodeoxglucose positron emission tomography
critical for accurate diagnosis, therapeutic strategy, and disease (PET) with CT (FDG PET/CT) is also used to investigate
prognosis [2]. potential adrenal cancer and provides anatomo-metabolic
There are multiple methods for discerning benign from information and has certain advantages over other techniques.
malignant lesions. The percutaneous biopsy is an eective It has faster attenuation and lower mismatches compared
METASTASIS DETECTION BY PET/CT 93

with PET alone, and malignancies may be detected earlier adrenal mass but did not include any peripheral areas so as to
since metabolic changes in the tissue may precede anatomical avoid partial volume eects. A similar sized ROI was chosen
changes [8]. Several studies have shown that FDG PET/CT on a non-aected liver region. The maximum standardized
high sensitivity, specicity and accuracy (all > 90%) for detect- uptake values (SUVmax) were calculated for both ROIs and the
ing adrenal metastatic disease [8-10]. ratio of the lesion to the normal tissue was calculated.
FDG PET/CT attributes make it not only useful for de- Statistical analysis. Age was expressed by mean with
termining if a mass is benign or cancerous but it also can be standard deviation (SD) and compared between groups by the
used as a non-invasive method to help characterize the cancer independent two samples t-test. Other categorical variables
itself which may facilitate the physician in diagnosis and were expressed by count with percentage, and the asso-
determination of treatment strategies. However, few studies ciations between categorical variables were tested with the
have investigated this issue. The objective of this study was Fishers exact test. The impact factors of adrenal metastasis
to use FDG PET/CT to characterize the properties of adrenal were expressed by their odds ratios with the 95% condence
lesions in patients with cancer. Specically, we investigated interval (CI) in the univariate and multivariate logistic re-
whether there are dierences between malignancies and gression models. The cuto point of SUV ratio (1.25) in the
benign lesions, are tumor characteristics associated with the predictions for adrenal metastasis was determined by the
location of the primary tumor, and are there predictors for Youdens index (the maximum of sensitivity+specicity-1)
adrenal metastasis. We also further evaluated the eective- in the ROC analysis. The variables with P-value less than
ness of FDG PET/CT in distinguishing between benign and 0.1 were included into the multivariate logistic regression
malignant lesions. models according to the selections by forward conditional
method. All statistical analyses were set with a signicance
Patients and methods level of 0.05 and performed using SPSS 15.0 statistics software
(SPSS Inc, Chicago, IL, USA).
This is a retrospective study of FDG PET/CT scan performed
from 2007 to 2010. Patients who visited the Chinese PLA Gen- Results
eral hospital either with or without a history of cancer prior to
the FDG PET/CT test were recruited into the study. This study FDG PET/CT examination was performed on 11570 sub-
was approved by the Institutional Review Board of Chinese PLA jects, including those with primary cancer (N = 3882) and
General Hospital, and the requirement of informed consent was those without cancer (N = 7688). Of all the patients examined,
waived due to the retrospective nature of this study. 374 subjects had adrenal gland lesions, and 3 young subjects
Study population. Eligible patients had an abnormal ad- (2 to 3 years of age) were excluded. The characteristics of the
renal lesion as determined by an initial PET/CT scan which adrenal lesions of remaining 371 subjects were analyzed (N
was diagnosed as adrenal metastasis (for those with known = 260 with a non-adrenal primary tumor and N =111 with
primary tumors), adrenal adenoma, or adrenal hyperplasia by unknown primary cancer).
a pathological follow-up (6 to 30 months following the initial Demographics and adrenal lesion characteristics. The 2
PET/CT). Patients were excluded from the study if they had groups of patients were similar with regard to age and gender
been treated for malignant or benign lesions in the adrenal (Table 1), and the majority of subjects in both groups were
gland, had diabetes, or any other disorder that may aect male. Over half (62.3%) the patients with a non-adrenal
glucose metabolism. primary tumor had a malignant adrenal lesion (Table 1). The
Image collection and analysis. All patients fasted for 4 types of non-adrenal primary tumors included lung (44.2%),
hours prior to FDG PET/CT. Images were acquired 60 minutes gastrointestinal (13.5%), and liver cancers (8.8%). On the
following the administration of 55.5 MBq/kg 18F-FDG using contrary, patients with unknown primary tumors did not have
the Siemens Biograph 64 HR (Siemens Medical Solutions malignant adrenal lesions; their lesions were either benign
USA, Molecular Imaging, Homan Estates, IL, USA). The CT (55.5%) or adrenal hyperplasia (45.0%) (Table 1).
scanning parameters were as follows: a low-dose CT (LDCT) Signicant dierences were observed in FDG PET/CT
scan was performed at 120 kV, 100 mAs, 0.8 s rotation, with characteristics between subjects with or without a known
a 1.25 mm slice width, and pitch of 0.9. The PET data were primary tumor. The proportion of patients with a tumor mass
acquired immediately after the CT scan in 3-D mode for 2.5 (in either the right or left adrenal) was small in both groups (<
min/bed and 3 or 7 dierent bed positions. The scan covered 8%) but was signicantly higher for the subjects with a known
from the bottom of the chin to the bottom of the pelvis. The primary tumor compared to those without (P-value 0.024)
PET images, including axial, sagittal and coronal images, were (Table 1). A similar proportion of patients in the 2 groups had
reconstructed by Fourier rebinning (FORE) ordered subset lesions that were characterized as nodules, enlarged adrenal
expectation maximization (OSEM) algorithm with attenua- glands, or normal (Table 1). More patients had normal right
tion correction. than left adrenals.
Image interpretation. The region of interest (ROI) was The SUVmax ratio was signicantly higher in subjects with
identied and was large enough to cover more than half the a known primary tumor than those without (Table 1). For
94 B. XU, J. GAO, L. CUI, H. WANG, Z. GUAN, S. YAO, Z. SHEN, J. TIAN

Table 1. Subject Demographics and Adrenal Lesion FDG PET/CT Characteristics.

With known primary tumor

Yes No P-value
(N = 260) (N = 111)

Demographics

Age (year) 61.7 13.3 63.3 12.4 0.280

Male
181 (69.6%) 80 (72.1%) 0.710
Abnormalities in adrenals
Adrenal metastasis or cancer 162 (62.3%) 0 (0.0%) <0.001*

Benign adrenal lesion 43 (16.5%) 61 (55.0%)

Adrenal hyperplasia 55 (21.2%) 50 (45.0%)

PET/CT characteristics Tumor mass 16 (6.2%) 1 (0.9%) 0.026*

Left adrenal Nodule 96 (36.9%) 50 (45.0%)

Enlarged 80 (30.8%) 40 (36.0%)

Normal 68 (26.2%) 20 (18.0%)

Tumor mass 20 (7.7%) 1 (0.9%) 0.024*

Right adrenal
Nodule 51 (19.6%) 17 (15.3%)

Enlarged 32 (12.3%) 16 (14.4%)

Normal 157 (60.4%) 77 (69.4%)

SUVmax ratio #
> 2.5 60 (26.3%) 2 (1.9%) <0.001*

2.5 103 (45.2%) 27 (25.2%)

absence of accumulated radioactivity 65 (28.5%) 78 (72.9%)

Unknown - 111 (100%)

Primary tumor Lung cancer 115 (44.2%) -

Gastrointestinal cancer 35 (13.5%) -

Liver cancer 23 (8.8%) -

Other 87 (33.5%) -
indicates a signicant association between the corresponding variable and group was observed. There were 36/371 (9.70) missing values in SUV.
* #

example, 26.3% and 1.9% of subjects with or without a known cancerous lesions) was greater for patients with adrenal me-
primary tumor, respectively, had a SUVmax ratio of > 2.5 (P < tastasis (44.4%) compared to those without adrenal metastasis
0.001); more patients with unknown a primary tumor (72.9% (1.1%; P < 0.001) (Table 2). A great proportion of patients with
versus 28.5%) had evidence of radioactive uorodeoxglucose adrenal metastasis had metastasized tumors in other parts of
accumulation (Table 1). their body (75.3% for patients with adrenal metastasis and
Associations between demographics and PET/CT char- 36.7% with no adrenal metastasis; P < 0.001).
acteristics in subjects with a primary tumor. Patients with About half the patients with adrenal metastasis (53.7%)
adrenal metastasis were younger than those whose primary had primary lung cancer, and less than 10% of patients had
tumor had not metastasized to the adrenal gland. (60.1 versus the primary tumors in the gastrointestinal track or the liver
64.4 years, respectively; P = 0.01) (Table 2). For both adrenals, (Table 2). Subjects with adrenal metastasis versus those with-
the proportion of patients with nodule or tumor mass was out had a higher frequency of primary lung tumors (53.7%
higher in those with a adrenal metastasis than in those without versus 28.6%, respectively; P 0.001) but a lower frequency
(Table 2). The proportion of patients with a SUVmax ratio of > of gastrointestinal cancer (9.3% versus 20.4%, respectively;
2.5 (the accepted value for distinguishing between benign and P = 0.014).
METASTASIS DETECTION BY PET/CT 95

For patients with primary tumors in the lung, gastrointes-

indicates the corresponding variable had a signicant inuence on renal metastasis. NA (not applicable) means the odds ratio was not performed due to small or zero count. #There were 32/260 (12.3%)
P-value

61.2 9.6 55.4 10.0 0.174

0.680
6 (54.5%) 11 (91.7%) 0.069
0.147

0.214
tinal track, or the liver, the frequency of either the left or right
adrenal having a nodule or tumor mass was greater in those

liver cancer (n=23)

For patients with
subjects with adrenal metastasis (Table 2), and this reached

6 (54.5%) 8 (66.7%)

1 (14.3%) 7 (58.3%)
6 (85.7%) 5 (41.7%)

8 (72.7%) 5 (41.7%)
3 (27.3%) 7 (58.3%)
0 (0.0%)
(n=12)
Adrenal metastasis
statistical signicance for patients with lung cancer. A higher

No
proportion of subjects with lung and gastrointestinal primary
tumors and adrenal metastasis had a SUVmax ratio > 2.5 than
the same group of patients without adrenal metastasis (P

0 (0.0%)
(n=11)
Yes
0.001). For the 3 subgroups of subjects, a higher proportion of
patients with adrenal metastasis also had metastasis in other
organs compared to those subjects without adrenal metastasis.
This was statistically signicant for subjects with lung and

For patients with gastrointestinal

P-value

58.2 12.7 68.4 10.4 0.017*

1 (7.7%) 13 (68.4%) 0.001*

*
0.481
0.700

14 (93.3%) 10 (50.0%) 0.009

gastrointestinal primary tumors (P 0.009). Patients with ad-
renal metastasis and gastrointestinal cancer were signicantly

Table 2. Summary for FDG PET/CT Characteristics of the adrenal Lesion in Subjects with Primary Tumors (n=260) and by Primary Tumor Type
younger than those without adrenal metastasis (mean age 58.2

cancer (n=35)

1 (6.7%) 10 (50.0%)
7 (46.7%) 6 (30.0%)
4 (26.7%) 4 (20.0%)

8 (61.5%) 6 (31.6%)
0 (0.0%)
(n=20)
Adrenal metastasis
versus 68.4 years, P = 0.017).

No
Summary of univariate analysis of predictors of adrenal
metastasis. The odds ratios from the univariate logistic regres-
sion models analysis indicated that the six variables of age, left

4 (30.8%)
(n=15)
Yes
and right adrenal FDG PET/CT characteristics, SUVmax ratio,
metastasis of the primary tumor to multiple locations, and
location of the primary tumor were signicant predictors of
adrenal metastasis (P 0.011) (Table 3). However, only age,
P -value

2 (2.7%) 18 (66.7%) <0.001*

*
67 (77.0%) 11 (39.3%) <0.001
0.123

0.127
0.630
SUVmax ratio, and the presence metastasis in multiple organs
were independent variables for adrenal metastasis according
lung cancer (n=115)
For patients with

to the selections by forward conditional method (Table 4).

62.4 13.0 66.5 11.6

20 (23.0%) 17 (60.7%)
44 (50.6%) 9 (32.1%)
23 (26.4%) 6 (21.4%)

36 (49.3%) 8 (29.6%)
35 (47.9%) 1 (3.7%)
(n=28)
Adrenal metastasis

There was one exception that age was excluded from the
No

multivariate analysis of lung cancer in Table 4 since it had no

independent impact on adrenal metastasis. For all subjects
after controlling for the SUVmax ratio and multiple metastatic
(n=87)
Yes

locations, the risk for developing adrenal metastasis decreased

slightly by each year of increasing age (OR = 0.97, P < 0.021).
Controlling for age and SUVmax ratio, subjects with metastasis
in multiple locations had a signicantly higher risk of adrenal
P -value

87 (53.7%) 28 (28.6%) <0.001*

7 (5.3%) 58 (61.1%) <0.001*

15 (9.3%) 20 (20.4%) 0.014*

36 (36.7%) <0.001
60.1 13.4 64.4 12.7 0.010*

Nodule or tumor mass 82 (50.6%) 30 (30.6%) 0.002*

Nodule or tumor mass 54 (33.3%) 17 (17.3%) 0.006*

11 (6.8%) 12 (12.2%) 0.176

49 (30.2%) 38 (38.8%) 0.176
metastasis than those with no additional metastasis (OR =
For all patients with known
primary tumor (n=260)

4.19, P < 0.001). After controlling for the other 2 variable,

67 (50.4%) 36 (37.9%)

40 (24.7%) 62 (63.3%)

subjects with a SUVmax ratio between > 0 and 2.5 or > 2.5
59 (44.4%) 1 (1.1%)
(n=98)
Adrenal metastasis

had signicantly higher risk of adrenal metastasis than those

No

who had no observable SUV (OR = 13.71, P < 0.001 and OR

= 461.54, P < 0.001, respectively).
SUVmax ratio and multiple metastasis locations were also
(n=162)

(75.3%)
122
Yes

independent variables for risk for adrenal metastasis in the

cohort of patients with lung cancer (Table 3). After controlling
for SUVmax ratio, the subjects with multiple metastatic sites
No adrenal radioactivity

Gastrointestinal cancer

versus those with no additional cancer sites had higher risk of

adrenal metastasis (OR = 4.12, P = 0.041). Similarly, patients
with SUVmax ratio > 0 and 2.5 as well as those with values >
accumulation

Primary tumor Lung cancer

Liver cancer

2.5 had a higher risk of adrenal metastasis than subjects with

no measurable SUV (OR = 33.85, P < 0.001 OR = 300.96, P <
PET/CT characteristics

Other
2.5
> 2.5

missing values in SUV.

Yes
No

0.001, respectively).
The variables of age, SUVmax ratio, and multiple metastatic
other positions
Metastasis in

sites were signicant predictors of adrenal metastasis in pa-

Right adrenal
Left adrenal
Age (year)

tients with gastrointestinal cancer. Only age and SUVmax ratio

were included in the multivariate analysis. After controlling
SUV

for age, the SUVmax ratio was found to be an independent

*
96 B. XU, J. GAO, L. CUI, H. WANG, Z. GUAN, S. YAO, Z. SHEN, J. TIAN

Table 3. Summary univariate logistic regression analysis.

All Cancer Subjects Subjects with
Subjects with Subjects with
gastrointestinal cancer
lung cancer (N = 115) liver cancer (N = 23)
(N = 260) (N = 35)
OR (95% CI) P-value OR (95% CI) P-value OR (95% CI) P-value OR (95% CI) P-value
Demographics
Age (year) 0.98 (0.96, 0.99) 0.011* 0.97 (0.94, 1.01) 0.142 0.92 (0.86, 0.99) 0.026* 1.07 (0.97, 1.17) 0.177
Gender Female - - - -
0.91 (0.05,
Male 0.58 (0.33, 1.03) 0.061 1.06 (0.43, 2.57) 0.905 0.31 (0.05, 1.95) 0.210 0.949
16.54)
PET/CT characteristic
Left adrenal Normal or enlarged - - - -
nodule or tumor mass 2.32 (1.37, 3.94) 0.002* 2.16 (0.88, 5.30) 0.093 2.04 (0.51, 8.23) 0.316 1.67 (0.31, 9.01) 0.553
Right adrenal Normal or enlarged - - - -
nodule or tumor mass 2.38 (1.29, 4.41) 0.006* 1.32 (0.47, 3.66) 0.596 1.45 (0.30, 7.09) 0.643 9.17 (0.86, 97.69) 0.066
SUVmax ratio 488.86 (58.30, 315.00 (26.73,
> 2.5 <0.001* <0.001* NA NA
4,098.89) 3,712.37)
2.5 15.42 (6.38, 37.28) <0.001* 40.50 (7.78, 210.79) <0.001* 17.33 (1.75, 171.66) 0.015* 8.40 (0.76, 93.34) 0.083
absence of radioactiv-
- - - -
ity accumulation
Metastasis in Yes 5.25 (3.05, 9.05) <0.001* 5.18 (2.09, 12.84) <0.001* 14.00 (1.54, 127.62) 0.019* 3.73 (0.65, 21.58) 0.141
other positions No - - - -
Primary Lung cancer 2.41 (1.32, 4.39) 0.004*
tumor Gastrointestinal cancer 0.58 (0.26, 1.28) 0.180
Liver cancer 0.71 (0.28, 1.79) 0.468
Other -
-
Reference group.*indicates the corresponding variable had a signicant inuence on adrenal metastasis. NA (not applicable) means the odds ratio was not
performed due to small or zero count.

Table 4. Summary for the impact factors of adrenal metastasis by multivariate logistic regression models.

For patients with

For all patients For patients with
gastrointestinal cancer
with primary tumor (n=260)a lung cancer (n=115) b
(n=35) c
OR (95% CI) P-value OR (95% CI) P-value OR (95% CI) P-value
Age (year) 0.97 (0.94, 0.99) 0.021* 0.91 (0.83, 1.00) 0.060
SUVmax ratio 300.96 (24.13,
> 2.5 461.54 (52.72, 4,040.50) <0.001* <0.001* NA
3,754.25)
< 2.5 13.71 (5.38, 34.92) <0.001* 33.85 (6.20, 184.68) <0.001* 20.92 (1.62, 270.40) 0.020*
absence of radioactivity
- - -
accumulation
Metastasis in other
Yes 4.19 (1.93, 9.13) <0.001* 4.12 (1.06, 15.98) 0.041*
positions
No - -
-
Reference group. *indicates the corresponding variable had a signicant inuence on adrenal metastasis. NA (not applicable) means the odds ratio
was not performed due to small or zero count. aP-value=0.321 in Hosmer and Lemeshow test indicates the multivariate model t well. bP-value=0.686
in Hosmer and Lemeshow test indicates the multivariate model t well. c P-value=0.136 in Hosmer and Lemeshow test indicates the multivariate
model t well

predictor of the risk of adrenal metastasis (OR = 20.93, P = Summary of multivariate analysis of predictors of adrenal
0.02). For subjects with liver cancer, no measured variable was metastasis. For patients with non-adrenal primary cancer,
signicant in the univariate analysis, and the sample size was multivariate analysis indicated that the SUVmax ratio, multiple
too small (N = 23) to perform multivariate analysis, therefore metastatic sites, and age were independent impact factors
they were not included in Table 4. of adrenal metastasis. Various combinations of the three
METASTASIS DETECTION BY PET/CT 97

Table 5. The sensitivity, specicity, PPV, and NPV of the predictions for adrenal metastasis by age, SUVmax ratio, and metastasis in other positions.

Adrenal metastasis
Sensitivity Specicity PPV NPV Accuracy
Yes No
For all patients with primary tumor
1. SUVmax ratio>0 Yes 126 37 77.8% 62.2% 77.3% 62.9% 71.9%
No 36 61
2. SUVmax ratio>1.25 Yes 113 12 69.8% 87.8% 90.4% 63.7% 76.5%
No 49 86
3. SUVmax ratio>1.25 or metastasis in other Yes 148 40 91.4% 59.2% 78.7% 80.6% 79.2%
positions No 14 58
4. SUVmax ratio>1.25 or metastasis in other Yes 155 56 95.7% 42.9% 73.5% 85.7% 75.8%
positions or age<55 years No 7 42
5. SUVmax ratio>1.25 or metastasis in other Yes 156 64 96.3% 34.7% 70.9% 85.0% 73.1%
positions or age<60 years No 6 34
6. SUVmax ratio>0 or metastasis in other Yes 153 55 94.4% 43.9% 73.6% 82.7% 75.4%
positions No 9 43
7. SUVmax ratio >0 or metastasis in other Yes 159 67 98.1% 31.6% 70.4% 91.2% 73.1%
positions or age<55 No 3 31
8. SUVmax ratio>0 or metastasis in other Yes 160 74 98.8% 24.5% 68.4% 92.3% 70.8%
positions or age<60 No 2 24
For the patients with lung cancer
9. SUVmax ratio>0 Yes 71 9 81.6% 67.9% 88.8% 54.3% 78.3%
No 16 19
10. SUVmax ratio>1.25 Yes 62 4 71.3% 85.7% 93.9% 49.0% 74.8%
No 25 24
11. SUVmax ratio>0 or metastasis in other Yes 84 15 96.6% 46.4% 84.8% 81.3% 84.3%
positions No 3 13
12. SUVmax ratio>1.25 or metastasis in other Yes 81 12 93.1% 57.1% 87.1% 72.7% 84.3%
positions No 6 16
For the patients with gastrointestinal cancer
13. SUVmax ratio >0 Yes 12 6 80.0% 70.0% 66.7% 82.4% 74.3%
No 3 14
14. SUVmax ratio >1.25 Yes 12 2 80.0% 90.0% 85.7% 85.7% 85.7%
No 3 18
15. SUVmax ratio>0 or age<55 Yes 13 7 86.7% 65.0% 65.0% 86.7% 74.3%
No 2 13
16. SUVmax ratio>0 or age<60 Yes 13 9 86.7% 55.0% 59.1% 84.6% 68.6%
No 2 11
17. SUVmax ratio>1.25 or age<55 Yes 13 3 86.7% 85.0% 81.3% 89.5% 85.7%
No 2 17
18. SUVmax ratio>1.25 or age<60 Yes 13 5 86.7% 75.0% 72.2% 88.2% 80.0%
No 2 15
PPV: positive predict value; NPV: negative predict value

variables were used to investigate predictive combinations In multivariate analysis of patients with lung cancer, the
for adrenal metastasis. The highest accuracy for predicting SUVmax ratio and metastasis in other positions were the two
adrenal metastasis was for the combination of SUVmax ratio independent impact factors for adrenal metastasis. The criteria
> 1.25 or metastasis in other positions with a 79.2%, 91.4% of an SUVmax ratio > 0 or metastasis in other positions had the
and 59.2% of accuracy, sensitivity and specicity, respectively highest sensitivity of 96.6%, lowest specicity of 46.4%, and
(Table 5). The combinations of the variables SUVmax ratio > 0 highest accuracy of 84.3%. For patients with gastrointestinal
or metastasis in other positions or age < 60 had the highest cancer, the criteria of an SUVmax ratio > 1.25 or age < 55 had
sensitivity of 98.8%, but a relative lower accuracy of 70.8% and the highest sensitivity of 86.7%, the second highest specicity
specicity of 24.5%. of 85.0%, and the highest accuracy of 85.7% (Table 5).
98 B. XU, J. GAO, L. CUI, H. WANG, Z. GUAN, S. YAO, Z. SHEN, J. TIAN

Discussion sensitivity, specicity and accuracy in predicting the presence

of adrenal metastasis.
This retrospective study used FDG PET/CT to evaluate The major limitation of this study is that it was a retro-
characteristics of adrenal lesions in patients with non-adrenal spective study. Also, our FDG PET/CT analysis was solely
primary tumor and with unknown cancer. It found that sub- quantitative and several studies have found that qualitative
jects with a known primary tumor were more likely to have analysis of FDG PET/CT may be more accurate in distinguish-
an adrenal tumor mass than subjects with unknown primary ing between malignant and benign adrenal lesions [10,12].
tumor. However, both cohorts of subjects had similar fre- The size of the study was small (N = 371) and subjects with
quency of other abnormalities including nodules and enlarged a primary lesion in the lung accounted for almost 50% of this
adrenals, suggesting these characteristics are not good predic- population. Hence, the ndings of the total population with
tors of the presence of adrenal lesion or adrenal cancer. primary tumors may predominately reect the characteristics
Prior studies have found that the use of the SUVmax ratio of the lung cancer population. Larger studies are necessary that
is highly accurate in distinguishing between benign and ma- include sizeable populations that represent multiple primary
lignant adrenal tumors [4,8-13]. These studies found that the tumor sites are needed to more fully assess the parameters that
sensitivity, specicity, and accuracy for distinguishing benign predict adrenal metastasis.
from cancerous tumors were between 85% to 100%. Similarly, In summary, FDG PET/CT was useful in characterizing adre-
this study found that FDG PET/CT was ecient at distin- nal lesions present in patients with and without known primary
guishing between benign and malignant adrenal cancers. It cancers. It identied variables that may useful in assessing the
found that about 40% of patients with primary tumors did not risk of developing adrenal metastasis. The use of FDG PET/CT
have a malignant adrenal lesion which is similar to what has may help facilitate the physician in diagnosis and determining
been previously described [1]. Moreover, the adrenal lesions the optimum treatment strategies in cancer patients.
in patients without cancer were all benign or due to adrenal
hyperplasia which is consistent with the low frequency (about Acknowledgment: We are in debt to Drs. Rui-Min Wang, Chang-
10%) of primary adrenal tumors [1]. Only 1.9% and1.1% of Bin Liu, Da-Yi Yin, Shu-Lin Yao of Department of Nuclear Medicine,
the subjects with no primary tumor or with a primary tumor Chinese PLA General Hospital for critical reading. This study was not
but no adrenal metastasis, respectively, had a SUVmax ratio > nancially supported by any institution and commercial sources.
2.5. In patients with a primary tumor, the left adrenal had
a higher frequency of a metastatic tumor mass than the left References
(50.6% versus 33.3%). The reasons for this are unclear.
The most common malignant lesions that metastasize to [1] MCLEAN K, LILIENFELD H, CARACCIOLO JT, HOFFE S,
the adrenal gland include malignant melanomas, kidney, liver, TOURTELOT JB et al. Management of isolated adrenal lesions
gastroinitestinal, breast, colon, lung, and bronchial carcino- in cancer patients. Cancer Control 2011; 2: 113-26.
[2] YOUNG WF The incidentally discovered adrenal mass. NEJM
mas [1]. In this study, the most common primary cancer that
2007; 356:601-9. http://dx.doi.org/10.1056/NEJMcp065470
metastasized to the adrenal gland was lung cancer (53.7%).
[3] UBEROI J, MUNVER R Surgical management of metastasis to
The majority of subjects with adrenal metastasis also had
the adrenal gland: open laparoscopic, and ablative approaches.
metastatic cancer in other body organs (75.3%) and this was
Curr Urol Rep 2009; 10: 67-72. http://dx.doi.org/10.1007/
even higher for the sub-group of patients with gastrointestinal s11934-009-0012-0
primary cancer (93.3%). These ndings are consistent with [4] LU Y, XIE D, HUAUNG W, GONG H, YU J 18F-FDG PET/CT
patients in this study who had adrenal metastasis having in the evaluation of adrenal masses in lung cancer patients.
advanced cancer. Neoplasm 2010; 57: 129-34. http://dx.doi.org/10.4149/neo_
Univariate analysis found that age, the SUVmax ratio and 2010_02_129
the presence of metastasized cancer in other organs were in- [5] CAOILI EM, KOROBKIN M, FRANCIS IR, COHAN RH,
dependent predictors for adrenal metastasis in patients with PLATT JF et al. Adrenal masses: characterization with com-
a non-adrenal primary tumor. The risk of metastasized adrenal bined unenhanced and delayed enhanced CT. Radiology 2002;
cancer from a primary tumor decreased with every increasing 222: 629-33. http://dx.doi.org/10.1148/radiol.2223010766
year of age. These same variables were independent predictors [6] BAE KT, FUANGTHARNTHIP P, PRASAD SR, JOE BN,
of adrenal metastasis in patients with primary tumors in the HEIKEN JP Adrenal masses: CT characterization with
lung and gastrointestinal track. Multivariate analysis found histogram analysis method. Radiology 2003; 228: 735-42.
that for patients with primary tumors that having a SUVmax http://dx.doi.org/10.1148/radiol.2283020878
ratio > 1.25 and the presence of metastasis in other organs [7] TSUSHIMA Y, ISHIZAKA H, MATSUMOTO M Adrenal
had a high sensitivity, specicity, and accuracy for predicting masses: dierentiation with chemical shift fast low-angle shot
adrenal metastasis. For patients with lung cancer the variables MR imaging. Radiology 1993; 186: 705-9
of the SUVmax ratio > 0 and multiple metastasis sited and for [8] OKADA M, SHIMONO T, KOMEYA Y, ANDO R, KAGAWA
patients with gastrointestinal cancer the variables of a SUVmax Y et al. Adrenal masses; the value of additional uorodeoxyglu-
ratio > 1.25 and age of < 55 years were the best combination of cose-positron emission tomography/computed tomography
METASTASIS DETECTION BY PET/CT 99

(FDG-PET/CT) in dierentiating between benign and ma- [11] ANSQUER C, SCIGLIANO S, MIRALLIE E, TAIEB D, BRU-
lignant lesion. Ann Nucl Med 2009; 23: 349-54. http://dx.doi. NAUD L et al. 18F-FDG PET/CT in the characterization and
org/10.1007/s12149-009-0246-4 surgical decision concerning adrenal masses: a prospective
[9] GRATZ S, KEMKE B, KAISER W, HEINIS J, BEHR TM et al. multicentre evaluation. Eur J Nucl Med Mol Imaging 2010;
Incidental non-secreting adrenal masses in cancer patients: 37: 1669-78. http://dx.doi.org/10.1007/s00259-010-1471-8
intra-individual comparison of 18F-uorodeoxyglucose [12] PARK BK, KIM CK, KIM B, CHOI JY. Comparison of delayed
positron emission tomography/computed tomography with enhanced CT and 18F-FDG PET/CT in the evaluation of adre-
computed tomography and shift magnetic resonance imaging. nal masses in oncology patients. J Comput Assist Tomogr 2007;
J Int Med Res 2010; 38: 633-44. 31: 550-6. http://dx.doi.org/10.1097/rct.0b013e31802fa8e1
[10] TESSONNIER L, SEBAG F, PALAZZO FF, COLAVOLPE C, [13] BOLAND GWL, BLAKE MA, HOLALKERE NS, HAHN PF.
DE MICCO C et al. Does 18F-FDG PET/CT add diagnostic PET/CT for the characterization of adrenal masses in patients
accuracy in incidentally identied non-secreting adrenal with cancer: qualitative versus quantitative accuracy in 150
tumours ? Eur J Ncul Med Mol Imaging. 2008; 35: 2018-25. consecutive patients. AJR 2009; 192: 956-62. http://dx.doi.
http://dx.doi.org/10.1007/s00259-008-0849-3 org/10.2214/AJR.08.1431