Prognostic impact of nomogram based on whole tumour size, tumour disappearance ratio on CT and SUVmax on PET in lung adenocarcinoma

Abstract

Objectives:
Lung adenocarcinoma frequently manifests as subsolid nodules, and the solid portion and ground-glass-opacity (GGO) portion on CT have different prognostic significance. Therefore, current T descriptor, defined as the whole tumour diameter without discrimination between solid and GGO, is insufficient. We aimed to determine the prognostic significance of solid tumour size and attempt to include prognostic factors such as tumour disappearance rate (TDR) on CT and SUVmax on PET/CT.

Methods:
Five hundred and ninety-five patients with completely resected lung adenocarcinoma were analyzed. We developed a nomogram using whole tumour size, TDR, and SUVmax. External validation was performed in another 102 patients.

Results:
In patients with tumours measuring ≤2 cm and >2 to 3 cm, disease free survival (DFS) was significantly associated with solid tumour size (P < 0.001), but not with whole tumour size (P = 0.052). Developed nomogram was significantly superior to the conventional T stage (area under the curve of survival ROC; P = 0.013 by net reclassification improvement) in stratification of patient survival. In the external validation group, significant difference was noted in DFS according to proposed T stage (P = 0.009).

Conclusions:
Nomogram-based T descriptors provide better prediction of survival and assessment of individual risks than conventional T descriptors.

Key points:
• Current measurement of whole tumour diameter including ground-glass opacity is insufficient • TDR enables differentiation between invasive solid portion and non-invasive GGO portion • SUVmax demonstrates the biological aggressiveness of the tumour • We developed a nomogram using whole tumour size, TDR, and SUVmax • Nomogram-based clinical T descriptors provide better prediction of survival.

Full text

CHEST
Prognostic impact of nomogram based on whole tumour size,
tumour disappearance ratio on CT and SUVmax on PET
in lung adenocarcinoma
So Hee Song
1
&Joong Hyun Ahn
2
&Ho Yun Lee
1
&Geewon Lee
3
&Joon Young Choi
4
&
Jun Kang
5
&Eun Young Kim
1
&Joungho Han
6
&O. Jung Kwon
7
&Kyung Soo Lee
1
&
Hong Kwan Kim
8
&Yong Soo Choi
8
&Jhingook Kim
8
&Young Mog Shim
8
Received: 7 May 2015 /Revised: 4 August 2015 /Accepted: 14 September 2015 /Published online: 11 October 2015
#European Society of Radiology 2015
Abstract
Objectives Lung adenocarcinoma frequently manifests as
subsolid nodules, and the solid portion and ground-glass-
opacity (GGO) portion on CT have different prognostic sig-
nificance. Therefore, current T descriptor, defined as the
whole tumour diameter without discrimination between solid
and GGO, is insufficient. We aimed to determine the prognos-
tic significance of solid tumour size and attempt to include
prognostic factors such as tumour disappearance rate (TDR)
on CT and SUVmax on PET/CT.
Methods Five hundred and ninety-five patients with complete-
ly resected lung adenocarcinoma were analyzed. We developed
a nomogram using whole tumour size, TDR, and SUVmax.
External validation was performed in another 102 patients.
Results In patients with tumours measuring ≤2 cm and
>2 to 3 cm, disease free survival (DFS) was significantly
associated with solid tumour size (P<0.001), but not
with whole tumour size (P=0.052). Developed nomo-
gram was significantly superior to the conventional T
stage (area under the curve of survival ROC; P=0.013
by net reclassification improvement) in stratification of
patient survival. In the external validation group, signifi-
cant difference was noted in DFS according to proposed T
stage (P=0.009).
Conclusions Nomogram-based T descriptors provide better
prediction of survival and assessment of individual risks than
conventional T descriptors.
Key points
•Current measurement of whole tumour diameter including
ground-glass opacity is insufficient
•TDR enables differentiation between invasive solid portion
and non-invasive GGO portion
So Hee Song and Joong Hyun Ahn contributed equally to this work.
Electronic supplementary material The online version of this article
(doi:10.1007/s00330-015-4029-0) contains supplementary material,
which is available to authorized users.
*Ho Yun Lee
hoyunlee96@gmail.com
*Geewon Lee
rabkingdom@naver.com
1
Department of Radiology and Center for Imaging Science, Samsung
Medical Center, Sungkyunkwan University School of Medicine, 50,
Ilwon-Dong, Gangnam-Gu, Seoul 135-710, Korea
2
Biostatistics Team, Samsung Biomedical Research Institute,
Seoul, Korea
3
Department of Radiology and Medical Research Institute, Pusan
National University Hospital, Pusan National University School of
Medicine, Busan, Korea
Eur Radiol (2016) 26:1538–1546
DOI 10.1007/s00330-015-4029-0
4
Departments of Nuclear Medicine, Samsung Medical Center,
Sungkyunkwan University School of Medicine, Seoul, Korea
5
Department of Pathology, Inchun St. Mary’s Hospital, College of
Medicine, Catholic University of Korea, Inchun, Korea
6
Department of Pathology, Samsung Medical Center, Sungkyunkwan
University School of Medicine, Seoul, Korea
7
Division of Respiratory and Critical Medicine of the Department of
Internal Medicine, Samsung Medical Center, Sungkyunkwan
University School of Medicine, Seoul, Korea
8
Department of Thoracic and Cardiovascular Surgery, Samsung
Medical Center, Sungkyunkwan University School of Medicine,
Seoul, South Korea

•SUVmax demonstrates the biological aggressiveness of the
tumour
•We developed a nomogram using whole tumour size, TDR,
and SUVmax
•Nomogram-based clinical T descriptors provide better pre-
diction of survival
Keywords Adenocarcinoma .Tumour staging .
Multidetector Computed Tomography .Positron-Emitted
Tomography .Lung neoplasm
Abbreviations
AIS Adenocarcinoma in situ
DFS Disease-free survival
FDG Fluorodeoxyglucose
GGO Ground-glass opacity
MIA Minimally invasive adenocarcinoma
NRI Net reclassification improvement
OS Overall survival
TDR Tumour disappearance rate
Introduction
Lung cancer is a heterogeneous and aggressive disease with
an overall five-year survival rate of 15 % [1,2].
Approximately 80 % of lung cancer patients are diagnosed
with adenocarcinoma, the most common histological subtype
[3]. Among patients with lung adenocarcinoma, some patients
have shown survival rates of 100 % after complete resection.
Those patients with excellent survival often have adenocarci-
noma in situ (AIS), minimally invasive adenocarcinoma
(MIA), and lepidic predominant adenocarcinomas according
to the new International Association for the Study of Lung
Cancer (IASLC)/American Thoracic Society (ATS)/European
Respiratory Society (ERS) classification [4]. On CT scans, these
indolent and less aggressive tumours of AIS, MIA, and lepidic
predominant adenocarcinomas frequently manifest as subsolid
lesions including a ground-glass opacity (GGO) component
[5].The GGO component of a subsolid tumour on lung window
setting of CT scans corresponds to the pathologic noninvasive
component in lung adenocarcinoma and is considered to have a
trivial effect on survival [5–9]. At mediastinal window images,
this GGO component disappears and previous investigators
have described this as the tumour disappearance rate (TDR).
Accordingtoseveralstudies,TDR seems to be inversely asso-
ciated with aggressiveness in lung cancer [10,11].
Current T descriptors of lung cancer are based on size,
invasion, and presence of separate tumour nodules, of which
the primary tumour size is measured as the longest visible
diameter, in other words, including GGO component [12,
13]. However, many studies have reported that the solid
tumour size, in other words, tumour size measurement of the
solid component only and excluding the GGO portion, more
effectively predicts tumour invasiveness and prognosis, indi-
cating that the current staging system possibly overestimates
the T status of lung adenocarcinomas [14–16]. However, those
related studies mostly included small tumours less than 3 cm
in size, whereas larger tumours are known to have a smaller
GGO component and more frequently present as totally solid
tumours [10,11,17].
Positron emission tomography (PET) is known to be highly
valuable for the differentiation of benign and malignant lung
lesions based on metabolic differences in glucose metabolism
[18]. Lepidic growth of lung cancer cells is represented by the
GGO component on CT scans and has low or no
fluorodeoxyglucose (FDG) uptake [19]. In contrast, pure solid
tumours, which exhibit a larger tumour size and higher max-
imum standardized uptake value (SUVmax) demonstrate
more pathological invasiveness and nodal metastasis [20].
Furthermore, SUVmax and solid tumour size have been re-
ported as independent prognostic factors influencing survival
in lung adenocarcinomas [16,20,21].
We hypothesized that the GGO component could act as a
confounding factor in early stage tumours, and, thus, eliminat-
ing the GGO component would lower the frequency of T
status overestimation. In addition, we assumed that using
SUVmax would reflect the pathological invasiveness of lung
tumours. Therefore, the aim of this study was to determine the
prognostic significance of solid tumour size and attempt to
include prognostic factors such as TDR from CT scans and
SUVmax from PET/CT for better clinical T description for the
development of an elaborate nomogram. External validation
of the results of our study was also performed.
Materials and methods
Approvals from the institutional review board at Samsung
Medical Center (IRB File number: 2014-01-074) and Pusan
National University Hospital (IRB number: E-2015084) was
obtained and informed consent was waived for reviewing the
patients’medical records.
Patients
We enrolled 990 patients with lung adenocarcinoma who
underwent complete resection at our institution (Samsung
Medical Center, Seoul, Korea) from September 2003 to
August 2011 (Supplementary Figure 1). Patients who had fac-
tors affecting prognosis (n=223) such as positive nodal stage,
history of neoadjuvant therapy, or concomitancy of other can-
cers were excluded. Patients who had CT images of inade-
quate quality (n=127; 124 with CT slice thickness of 3 mm
or more and three patients with combined atelectasis/
Eur Radiol (2016) 26:1538–1546 1539

pneumonia) or insufficient pathologic specimen (n=45) for
tumour evaluation were also excluded. Accurate tumour size
measurement is essential for our study, thus we excluded pa-
tients with inadequate CT images, which precluded precise
measurement, and only included patients with a CT slice
thickness of 2.5 mm or less. Thus, our final study group in-
cluded 595 patients with thin-section CT and preoperative
PET/CT scans. Surgery was performed according to tumour
extent and underlying condition. Sublobar resections such as
wedge resection were allowed if the tumour appeared as a
nonsolid nodule on CT and had no lymph node metastasis
based on intraoperative assessment. Electronic medical re-
cords were searched by one of the authors (S.H. S.) for date
of documented recurrence, last follow-up, or death.
External validation of the results of our study was per-
formed using an independent dataset of 102 patients with
completely resected lung adenocarcinomas from a different
tertiary referral institute in a different district (Pusan
National University Hospital, Busan, Korea) during the same
time period (September 2003 to August 2011). All patients in
the externalvalidation group had undergoneboth preoperative
thin-section CT and PET/CT scans. One of the authors (G.L.)
searched the electronic medical records for date of document-
ed recurrence, last follow-up, or death.
Imaging and pathologic analysis
All patients underwent contrast-enhanced helical CT with a
16-detector row (LightSpeed 16, GE Healthcare) scanner.
Section thickness ranged from 1-2.5 mmfor transverse images
with both mediastinal (width, 400 HU; level, 20 HU) and lung
(width, 1500 HU; level, -700 HU) window settings available.
One chest radiologist, unaware of clinical information,
assessed the CT scans. GGO was defined as a misty increase
in lung attenuation that did not obscure the underlying vascu-
lar markings [22]. All patients had both axial and coronal CT
images available for evaluation and the longest tumour diam-
eter was measured on either axial or coronal CT images where
the largest tumour dimension appeared. The radiologist mea-
sured the whole tumour size including GGO on lung windows
and solid tumour size was defined as the maximum dimension
of the solid component of the mediastinal window excluding
GGO. From these measurements, TDR was calculated [23].
TDR (%) was defined as [1-(tumour area of the mediastinal
windows/tumour area of the lung windows)] × 100 [17,24].
PET/CT scans were acquired as previously reported [25].
Before PET/CT examination, all patients fasted for at least
6 hours and after a normal blood glucose level in peripheral
blood was ensured, patients received an intravenous injection
of 379 MBq (10 mCi) of fluorine-18-FDG and rested for ap-
proximately 45 minutes before scanning. Scans were acquired
with a PET/CT device (Discovery LS; GE Medical Systems,
Milwaukee, WI). For a semiquantitative analysis of FDG
uptake, a nuclear medicine physician, unaware of clinical in-
formation, placed regions of interest over the most intense
area of FDG accumulation, which was interpreted as
SUVmax. Both CT and PET scans were performed within
25 days of each other. The time between CT and surgical
resection ranged from 1 to 28 days (mean, 20 days) and the
interval between PET and surgical resection ranged from 1 to
28 days (mean, 25 days).
In the external validation group, contrast-enhanced helical
CT was obtained with either a 4-detector (Lightspeed Plus, GE
Healthcare) or 16-detector row (Sensation 16, Siemens) scan-
ner. Section thickness ranged from 1 to 2.5 mm for transverse
images with both mediastinal (width, 400 HU; level, 20 HU)
and lung (width, 1500 HU; level, -700 HU) window settings
available.Measurements for whole and solid tumour size were
performed identically to the training group. PET/CT scans
were acquired (Gemini, Philps, Milpitas, CA) and the
SUVmax of each tumour was obtained. Similar to the training
group, both CT and PETscans were performed within 23 days
(mean, 8 days) of each other. The time between CT and sur-
gical tumour removal ranged from 1 to 28 days (mean,
10 days) and the interval between PET and surgical tumour
removal ranged from 1 to 28 days (mean, 15 days).
Each resection specimen was evaluated by a pathologist
with 18 years of experience in pulmonary pathology with
standard pathologic methods [26]. Comprehensive histologi-
cal subtyping was performed semi-quantitatively to the
nearest 5 % level, and the most predominant subtype was
determined [27]. Tumours were stratified into the following
three grades based on histologically predominant subtype:
low grade (AIS, MIA, and lepidic predominant); intermediate
grade (papillary and acinar predominant); and high grade
(micropapillary and solid predominant) [9,25].
Statistical analysis
Differences in TDR value and SUVmax among four sub-
groups according to whole tumour size were assessed by the
Kruskal-Wallis test and post-hoc test of Tukey. Overall sur-
vival (OS) was defined as the time from surgery to death from
any cause or to the date of the last visit of patients without
events. Disease-free survival (DFS) was defined as the time
from surgery to the first event of recurrence or the last follow-
up visit. Recurrence was defined as any documented clinical
or pathologic evidence of local or distant disease recurrence.
OS and DFS curves were estimated using the Kaplan–Meier
method and compared using the log-rank test. The association
between DFS or OS and the following clinicopathologic var-
iables were estimated: age, sex, whole tumour size/solid tu-
mour size, TDR, SUVmax, and histological subtypes and
grades of lung adenocarcinomas based on the IASLC/ATS/
ERS classification scheme. Univariate and multivariate Cox
1540 Eur Radiol (2016) 26:1538–1546

regression analyses were performed to investigate the effects
of clinical variables on survival rates.
Next, we developed an elaborative nomogram based on the
results of the Cox regression analysis, which contained the
three factors of whole tumour size, TDR, and SUVmax. To
evaluate the discriminative power of the new clinical T de-
scriptors, we calculated the receiver operating characteristic
(ROC) curve and used the net reclassification improvement
(NRI) method, which evaluates the proportion of patients
moving accurately or inaccurately from one risk category to
another after applying the nomogram [28]. The nomogram
predicts three-year DFS after curative resection, and external
validation of the nomogram was carried out in a different
patient group by estimating OS and DFS curves using the
Kaplan–Meier method and log-rank test. Statistical Package
for the Social Sciences (SPSS) software (version 21; SPSS,
Inc., Chicago, IL, USA) was used for statistical analysis. The
level of statistical significance was set at P<0.05.
Results
Clinical features and distributions of TDR value
and SUVmax
Characteristics of 595 patients are summarized in Table 1.
Supplementary Table 1demonstrates the differences in tu-
mour sizes when considering whole tumour size or solid tu-
mour size. Distributions of TDR value and SUVmax accord-
ing to whole tumour size are summarized in Table 2.TDR
value and SUVmax were significantly different among the
four groups (P<0.001 for both TDR and SUVmax). As the
whole tumour size increased, TDR value tended to decrease,
which indicates a decreasing GGO component, while
SUVmax tended to increase.
Survival analyses
Median follow-up period after surgery was 7.95 years (95 %
CI 7.65-8.24), during which tumours recurred in 100 patients
(16.8 %). Survival analysis according to whole and solid tu-
mour sizes is summarized in Supplementary Table 2. DFS was
not significantly different between patients with whole tumour
size≤2 cm and those with a whole tumour size of >2 to 3 cm
Table 1 Characteristics of participants (N=595)
Characteristics Number of patients
Age, years, mean± SD 60.1±10
Sex
Male 286 (48.1)
Smoker
Current or ex-smoker 214(35.9)
Never-smoker 381(64.0)
Performance score
0 295(49.6)
1 289(48.6)
2 11(1.8)
Tumor size on C T, cm
Whole tumor size
≤2 254 (42.7)
>2 to 3 216 (36.3)
>3 to 5 107 (18.0)
>5 18 (3.0)
Solid tumor size
≤2 389 (65.4)
>2 to 3 134 (22.5)
>3 to 5 64 (10.8)
>5 8 (1.3)
TDR, mean±SD 55.1±35
SUVmax, mean± SD 4.4± 4
Lymphatic in vas ion
Positive 113 (19)
Vascular invasion
Positive 19 (2.5)
Pleural invasion
Positive 121 (20.3)
Perineural invasion
Positive 7 (1.2)
Lymph node metastases
Positive 0
Procedure
Wedge resection 97 (16.3)
Lobectomy 494 (83.0)
Pneumonectomy 4 (0.7)
Tumor differentiation
Well-differentiated 278 (46.7)
Moderately-differentiated 249 (41.8)
Poorly-differentiated 68 (11.4)
Tumor type
AIS 39 (6.6)
MIA 33 (12.1)
Lepidic 133 (22.4)
Acinar 260 (43.7)
Papillary 48 (8.1)
Micropapillary 13 (2.2)
Solid 59 (9.9)
Table 1 (continued)
Characteristics Number of patients
Variant 10 (1.7)
Numbers in parentheses are percentages. TDR tumor disappearance rate;
SUVmax maximum standardized uptake value, AIS adenocarcinoma in
situ; MIA minimally invasive adenocarcinoma; SD standard deviation
Eur Radiol (2016) 26:1538–1546 1541

(P=0.052; Fig. 1a). However, a significant difference in DFS
was noted between patients with a solid tumour size≤2cm
and those with a solid tumour size of >2 to 3 cm (P<0.001;
Fig. 1b). In addition, there were significant differences in DFS
between patients with whole tumour size >2 to 3 cm and those
with whole tumour size of >3 to 5 cm (P<0.001),andbetween
patients with solid tumour size >2 to 3 cm and those with solid
tumour size of >3 to 5 cm (P=0.002). However, DFS was not
statistically significant between tumours of >3 to 5 cm and
those with >5 cm according to either whole tumour size (P=
0.274) or solid tumour size (P=0.943).
Univariate analysis of OS and DFS is demonstrated in
Supplementary Table 3. Cox multivariate survival analysis
(Table 3) revealed that age (P=0.019), sex (P=0.006), whole
tumour size (P=0.048), histological grade (P=0.003), and
SUVmax (P<0.001) were independently associated with
OS. Meanwhile, whole tumour size (P=0.049), histological
grade (P=0.009), and TDR (P= 0.001) were independently
associated with DFS.
Consequently, a nomogram (Fig. 2) for predicting the prob-
ability of three-year DFS was developed using the three var-
iables of whole tumour size, TDR, and SUVmax based on
these patients. Each point can be established by drawing a line
from each variable to the point axis, and total points are
calculated as the sum of the three points of whole tumour size,
TDR, and SUVmax, thus indicating the T category.
When we newly classified patients according to this new T
classification, there were significant differences (overall
P<.001) in survival between all categories: proposed T1a ver-
sus proposed T1b, P=0.013; proposed T1b versus proposed
T2a, P=0.027; and proposed T2a versus proposed T2b,
P<.001 (Fig. 1c). In particular, DFS was significant
(P<.001) between proposed T2a and proposed T2b, which
was not statistically significant according to either whole tu-
mour size (P=0.274) or solid tumour size (P=0.943).
Discrimination of the developed nomogram was significantly
superior to that of the conventional T classification (area under
the curve of survival ROC, 0.773 vs. 0.675; P=0.013 by net
reclassification improvement).
Supplementary Figure 2is a waterfall plot demonstrating
distribution in T status from current T category to the proposed
T category. There were 204, 182, 94, and 14 patients with T1a,
T1b, T2a, and T2b tumours, respectively, according to the
current stage. When reclassified according to the proposed T
stage, 152, 85, 180, 77 patients had T1a, T1b, T2a, and T2b
tumours, respectively.
External validation
The external validation study group included 102 patients (49
men and 53 women; mean age of 60.0±9 years) from another
institute in a different district. Comparison of demographics
between the development group and external validation group
are demonstrated in Supplementary Table 4.Among102pa-
tients, 31 (30 %) demonstrated tumour recurrence, and nine
(9 %) died. We performed survival analyses according to the
conventional T classification and then assigned a new T status
with reference to the nomogram.
Results of survival analysis according to the current and
proposed T category are shown in Fig. 3. For DFS, a
Table 2 Distribution of TDR value and SUVmax according to whole
tumor size
Whole tumor size
Tot al ≤2
(n=204)
>2 to 3
(n= 182)
>3 to 5
(n= 94)
>5
(n=14)
Pvalue
TDR, mean±SD 73.2±31 45.4±30 32.3± 27 21.3±20 <0.001
SUVmax,
mean±SD
1.9± 2 4.9± 3 8.1 ±11 13.4±8 <0.001
TDR tumor disappearance rate; SD standard deviation; SUVmax maxi-
mum standardized uptake value
Fig. 1 Disease-free survival analysis according to awhole tumour size, bsolid tumour size, and cproposed T stage based on the nomogram
1542 Eur Radiol (2016) 26:1538–1546

significant difference was noted according to the proposed T
category (P=0.009), whereas there was no significant differ-
ence according to the conventional T classification (P=0.120).
In terms of OS, although no statistical significance was noted
in either stage, survival curves clearly show more precise
stratification of patients according to the proposed T stage.
Furthermore, there were no deaths in patients in the T1a and
T1b groups according to the proposed T classification.
Discussion
T status of lung cancer is critical to therapeutic decision-mak-
ing, and according to the current staging classification of lung
cancers, it is determined by the greatest dimension of the pri-
mary tumour. For example, a 2.2 cm-sized entirely solid lung
adenocarcinoma and a 2.2 cm-sized part-solid (with 1 cm of
solid portion) lung adenocarcinoma are both defined as T1b
(Supplementary Figure 3a and 3b), regardless of the internal
composition. However, adenocarcinomas frequently manifest
as subsolid tumours composed of both solid and GGO com-
ponents. The GGO component on CT scans reflect the patho-
logic noninvasive component, representing the lepidic growth
pattern of adenocarcinoma [5,6,8]. Therefore, there has been
increasing interest in the current T classification using whole
tumour size including GGO component for predicting the prog-
nosis of patients. Tsutani et al. [16] reported that solid tumour
size on HRCT is more useful for predicting pathologic tumour
invasiveness than whole tumour size, and Nakamura et al. [15]
reclassified tumours according to solid tumour size, revealing
better classification of patient prognosis in early lung cancer.
However, by estimating the solid tumour only, it is impossible
todiscriminatebetweena2.2cm-sizedpart-solid(with1cm
of solid portion) lung adenocarcinoma from a 1 cm-sized
completely solid tumour (Supplementary Figure 3b and 3c).
In this regard, TDR, which provides distinction between solid
portion and GGO, has been shown to be a strong indicator of
tumour invasiveness [10,14,24,29–31].
These considerations led us to the investigation ofa method
to distinguish among all three tumours (Supplementary
Figure 3a,3b,and3c) and reflect the aggressiveness of the
tumour. Thus, we aimed to create a lung adenocarcinoma-
specific T status criteria, which was a strong predictor of sur-
vival and could be measured from standard preoperative im-
aging based variables only. The concept of our developed
nomogram is based on the three components of TDR, whole
tumour size, and SUVmax. TDR enables differentiation be-
tween pathologic invasive solid portion and non-invasive
GGO portion. In addition, FDG uptake is proportional to the
metabolic activity of viable tumour cells, which demonstrates
Table 3 Multivariate analysis for overall survival and disease-free
survival
Characteristic HR 95 % CI Pvalue
OS
Age 1.07 1.01-1.13 0.019
*
Sex 0.24 0.09-0.66 0.006
*
Whole tumor size 11.36 1.02-126.85 0.048
*
Histologic grade 6.35 1.85-21.83 0.003
*
TDR 0.99 0.97-1.01 0.411
SUVmax 1.04 1.02-1.06 <0.001
*
DFS
Age 1.01 0.99-1.04 0.332
Sex 0.69 0.45-1.05 0.080
Whole tumor size 1.90 1.00-3.60 0.049
*
Histologic grade 3.32 1.36-8.09 0.009
*
TDR 0.98 0.97-0.99 0.001
*
SUVmax 1.02 0.99-1.04 0.172
*
Statistically significant at P<.05. HR hazard ratio; CI confidence inter-
val; OS overall survival; DFS disease-free survival; TDR tumor shadow
disappearance rate; SUVmax maximum standardized uptake value
Fig. 2 Nomogram based on a score system as new clinical T descriptors with lung adenocarcinoma
Eur Radiol (2016) 26:1538–1546 1543

the biological aggressiveness of the tumour [32,33].
However, current T category is defined in terms of the
size, location, and regional invasion of the primary
tumour [34]. Therefore, for accurate evaluation of the
tumour, we thought that addition of FDG uptake into the
T category, along with the traditional anatomic informa-
tion, could provide better prognostic stratification of
patients with lung cancer. We also included the whole
tumour size, which is in contrast to previous studies about
modified T staging, which did not consider the whole
tumour size at all [15,16]. However, whole tumour size
is the basis and foundation of current T status, which was
established throughout previous invaluable studies, and
we believe that including whole tumour size is indeed
mandatory in the staging process [35].
Our results are consistent with the aforementioned
studies in that solid tumour size more clearly classified
DFS than did whole tumour size. However, a general
weak point of previous studies suggesting solid tumour
size as a T factor is that they were based on small
tumours [10,14,24,29–31]. We included not only ear-
ly lung adenocarcinomas, but also large tumours up to
7 cm in size. With regard to our results, classification
according to solid tumour size or TDR is useful in early
tumours but is insufficient in larger tumours of T2
status.
Furthermore, PET/CT is already part of the standard
protocol for preoperative assessment of lung cancer, and
results of our study corroborate those of previous studies
in that SUVmax successfully predicted tumour invasive-
ness and was an independent factor of survival [16,21].
In particular, DFS was significantly different between pro-
posed T2a versus proposed T2b, but not significantly
different according to whole tumour size and solid tumour
size in our study. This implies that SUVmax carries more
clinical significance in the prognosis of larger tumours
than with smaller tumours. More importantly, given that
SUVmax is known to correlate well with tumour
Fig. 3 Survival curves of patients in the external validation group according to the current T stage and new T stage
1544 Eur Radiol (2016) 26:1538–1546

differentiation and histological subtype, further prognostic
discrimination may be feasible, even among pure solid
tumours of the same size, by using SUVmax [25,36].
The basis of our proposed T classification is that, in
early stages of adenocarcinoma, the GGO component
acts as a confounding factor that negatively influences
tumour invasiveness and aggressiveness, whereas in
higher stages, using both the entire tumour size along
with SUVmax results in effective patient stratification.
Considering the clinical significance of the developed
nomogram, our attempt to combine whole tumour size,
GGO portion (namely TDR), and SUVmax, is unique in
nature and superior to previously suggested T classifica-
tions in that it can be applied to both small and large
tumours [15,16].
In addition, after adjustment for age, sex, histological
grade, and whole tumour size, SUVmax was an indepen-
dent prognostic factor for DFS, whereas TDR was an
independent prognostic factor for OS. Therefore, we
conclude that our proposed T category, which considers
whole tumour size, TDR, and SUVmax, demonstrates bet-
ter stratification of patients with clinical T1a-T2bN0M0
lung adenocarcinoma than does the current T classifica-
tion. The efficacy of our staging system was externally
validated in a patient dataset from another institute in a
different district.
Our study had several limitations. First, as is typical of
retrospective studies, our study was limited by biases such
as lack of random assignment, patient selection, and
incomplete data acquisition. The use of patients from a
single large cancer centre may reflect a cohort with more
aggressive disease, which is the referral pattern for such
centres. However, we attempted to overcome this limita-
tion through external validation of patients from another
tertiary cancer centre. Third, although our study group
covered adenocarcinomas of clinical T1a-T2b stages, the
majority of patients were of stages T1a and T1b, and the
number of patients with large tumours (>5 cm) was
relatively low. Finally, CT scans were performed using
various vendors, which might have influenced the CT
quality. However, we strictly included only patients with
thin-section CT images.
In conclusion, TDR was more effective in predicting
tumour invasiveness and OS than the current T stage
system. SUVmax was also a useful factor for predicting
pathologic high-grade malignancy and DFS. Furthermore,
the elaborate nomogram we developed, which is based on a
combination of whole tumour size, TDR, and SUVmax,
accurately predicted survival in the external validation group.
Therefore, we suggest that our proposed T classification
demonstrates more effective stratification of patient survival
and may lead to more adequate patient treatment than the
current T classification.
Acknowledgments The scientific guarantor of this publication is Ho
Yun Lee. The authors of this manuscript declare no relationships with any
companies, whose products or services may be related to the subject
matter of the article. The authors state that this work has not received
any funding. One of the authors has significant statistical expertise. Insti-
tutional Review Board approval was obtained. Written informed consent
was waived by the Institutional Review Board. Methodology: retrospec-
tive, diagnostic or prognostic study, performed at one institution.
References
1. Aberle DR, Adams AM, Berg CD et al (2011) Reduced lung-cancer
mortality with low-dose computed tomographic screening. N Engl J
Med 365:395–409
2. Jemal A, Siegel R, Xu J, Ward E (2010) Cancer statistics, 2010. CA
Cancer J Clin 60:277–300
3. Chen F, Cole P, Bina WF (2007) Time trend and geographic pat-
terns of lung adenocarcinoma in the United States, 1973-2002.
Cancer Epidemiol Biomarkers Prev 16:2724–2729
4. Russell PA, Wainer Z, Wright GM, Daniels M, Conron M, Williams
RA (2011) Does lung adenocarcinoma subtype predict patient sur-
vival?: A clinicopathologic study based on the new International
Association for the Study of Lung Cancer/American Thoracic
Society/European Respiratory Society international multidisciplin-
ary lung adenocarcinoma classification. J Thorac Oncol 6:1496–
1504
5. Kuriyama K, Seto M, Kasugai T et al (1999) Ground-glass opacity
on thin-section CT: value in differentiating subtypes of adenocarci-
noma of the lung. AJR Am J Roentgenol 173:465–469
6. Aoki T, Tomoda Y, Watanabe H et al (2001) Peripheral lung ade-
nocarcinoma: correlation of thin-section CT findings with histolog-
ic prognostic factors and survival. Radiology 220:803–809
7. Higashiyama M, Kodama K, Yokouchi H et al (1999) Prognostic
value of bronchiolo-alveolar carcinoma component of small lung
adenocarcinoma. Ann Thorac Surg 68:2069–2073
8. Lee HY, Lee KS (2011) Ground-glass opacity nodules: histopathol-
ogy, imaging evaluation, and clinical implications. J Thorac
Imaging 26:106–118
9. Yoshizawa A, Motoi N, Riely GJ et al (2011) Impact of proposed
IASLC/ATS/ERS classification of lung adenocarcinoma: prognos-
tic subgroups and implications for further revision of staging based
on analysis of 514 stage I cases. Mod Pathol 24:653–664
10. Okada M, Nishio W, Sakamoto T et al (2004) Correlation between
computed tomographic findings, bronchioloalveolar carcinoma
component, and biologic behavior of small-sized lung adenocarci-
nomas. J Thorac Cardiovasc Surg 127:857–861
11. Okada M, Nishio W, Sakamoto T, Uchino K, Tsubota N (2003)
Discrepancy of computed tomographic image between lung and
mediastinal windows as a prognostic implication in small lung ad-
enocarcinoma. Ann Thorac Surg 76:1828–1832, discussion 1832
12. Austin JH, Garg K, Aberle D et al (2013) Radiologic implications
of the 2011 classification of adenocarcinoma of the lung. Radiology
266:62–71
13. Greaves SM, Brown K, Garon EB, Garon BL (2011) The new
staging system for lung cancer: imaging and clinical implications.
J Thorac Imaging 26:119–131
14. Murakawa T, Konoeda C, Ito T et al (2013) The ground glass
opacity component can be eliminated from the T-factor assessment
of lung adenocarcinoma. Eur J Cardiothorac Surg 43:925–932
15. Nakamura S, Fukui T, Taniguchi Tet al (2013) Prognostic impactof
tumor size eliminating the ground glass opacity component: modi-
fied clinical T descriptors of the tumor, node, metastasis classifica-
tion of lung cancer. J Thorac Oncol 8:1551–1557
Eur Radiol (2016) 26:1538–1546 1545

16. Tsutani Y, Miyata Y, Nakayama H et al (2013) Solid tumor size on
high-resolution computed tomography and maximum standardized
uptake on positron emission tomography for new clinical T descrip-
tors with T1 lung adenocarcinoma. Ann Oncol 24:2376–2381
17. Okada M, Koike T, Higashiyama M, Yamato Y, Kodama K,
Tsubota N (2006) Radical sublobar resection for small-sized non–
small cell lung cancer: A multicenter study. J Thorac Cardiovasc
Surg 132:769–775
18. Hubner KF, Buonocore E, Singh SK, Gould HR, Cotten DW
(1995) Characterization of chest masses by FDG positron emission
tomography. Clin Nucl Med 20:293–298
19. Yap CS, Schiepers C, Fishbein MC, Phelps ME, Czernin J (2002)
FDG-PET imaging in lung cancer: how sensitive is it for bronchi-
oloalveolar carcinoma? Eur J Nucl Med Mol Imaging 29:1166–
1173
20. Tsutani Y, Miyata Y, Nakayama H et al (2014) Segmentectomy for
clinical stage IA lung adenocarcinoma showing solid dominance on
radiology. Eur J Cardiothorac Surg 46:637–642
21. Tsutani Y, Miyata Y, Nakayama H et al (2012) Prognostic signifi-
cance of using solid versus whole tumor size on high-resolution
computed tomography for predicting pathologic malignant grade
of tumors in clinical stage IA lung adenocarcinoma: a multicenter
study. J Thorac Cardiovasc Surg 143:607–612
22. Lee HY, Choi YL, Lee KS et al (2014) Pure ground-glass opacity
neoplastic lung nodules: histopathology, imaging, and manage-
ment. AJR Am J Roentgenol 202:W224–W233
23. Lee HY, Han J, Lee KS et al (2009) Lung adenocarcinoma as a
solitary pulmonary nodule: prognostic determinants of CT, PET,
and histopathologic findings. Lung Cancer 66:379–385
24. Nakayama H, Yamada K, Saito H et al (2007) Sublobar resection
for patients with peripheral small adenocarcinomas of the lung:
surgical outcome is associated with features on computed tomo-
graphic imaging. Ann Thorac Surg 84:1675–1679
25. Lee HY, Jeong JY, Lee KS et al (2013) Histopathology of lung
adenocarcinoma based on new IASLC/ATS/ERS classification:
prognostic stratification withfunctional and metabolic imaging bio-
markers. J Magn Reson Imaging 38:905–913
26. Lester SC (2006) Manual of surgical pathology, 2nd edn. Elsevier
Churchill Livingstone, New York
27. Travis WD, Brambilla E, Noguchi M et al (2011) International
association for the study of lung cancer/american thoracic
society/european respiratory society international multidisci-
plinary classification of lung adenocarcinoma. J Thorac Oncol
6:244–285
28. Vickers AJ, Pepe M (2014) Does the net reclassification improve-
ment help us evaluate models and markers? Ann Intern Med 160:
136–137
29. Hashizume T, Yamada K, Okamoto N et al (2008) Prognostic sig-
nificance of thin-section CT scan findings in small-sized lung ade-
nocarcinoma. Chest 133:441–447
30. Matsuguma H, Yokoi K, Anraku M et al (2002) Proportion of
ground-glass opacity on high-resolution computed tomography in
clinical T1 N0 M0 adenocarcinoma of the lung: a predictor of
lymph node metastasis. J Thorac Cardiovasc Surg 124:278–284
31. Takahashi M, Shigematsu Y, Ohta M, Tokumasu H, Matsukura T,
Hirai T (2014) Tumor invasiveness as defined by the newly pro-
posed IASLC/ATS/ERS classification has prognostic significance
for pathologic stage IA lung adenocarcinoma and can be predicted
by radiologic parameters. J Thorac Cardiovasc Surg 147:54–59
32. Downey RJ, Akhurst T, Gonen M et al (2004) Preoperative F-18
fluorodeoxyglucose-positron emission tomography maximal stan-
dardized uptake value predicts survival after lung cancer resection. J
Clin Oncol 22:3255–3260
33. Higashi K, Ueda Y, Arisaka Y et al (2002) 18F-FDG uptake as a
biologic prognostic factor for recurrence in patients with surgically
resected non-small cell lung cancer. J Nucl Med 43:39–45
34. van Rens MT, de la Riviere AB, Elbers HR, van Den Bosch JM
(2000) Prognostic assessment of 2,361 patients who underwent
pulmonary resection for non-small cell lung cancer, stage I, II,
and IIIA. Chest 117:374–379
35. Goldstraw P, Crowley J, Chansky K et al (2007) The IASLC Lung
Cancer Staging Project: proposals for the revision of the TNM stage
groupings in the forthcoming (seventh) edition of the TNM
Classification of malignant tumours. J Thorac Oncol 2:706–714
36. Kadota K, Colovos C, Suzuki K et al (2012) FDG-PET SUVmax
combined with IASLC/ATS/ERS histologic classification improves
the prognostic stratification of patients with stage I lung adenocar-
cinoma. Ann Surg Oncol 19:3598–3605
1546 Eur Radiol (2016) 26:1538–1546