Small (<20 mm) Enhancing Hepatic Nodules Seen on

Arterial Phase MR Imaging of the Cirrhotic Liver:

Clinical Implications
Yong Yeon Jeong1,2, Donald G. Mitchell1 and Tamotsu Kamishima1,3
1
Department of Radiology, Thomas Jefferson University Hospital, 132 S. 10th St., 1096 Main Bldg.,
Philadelphia, PA 19107.
2
Present address: Department of Diagnostic Radiology, Chonnam National University Medical
School, 8 Hackdong, Dongku, Kwangju, South Korea.
3
Present address: Department of Radiology, Hokkaido University School of Medicine, North-15,
West-7, Kita-Ku, 0608638, Sapporo, Japan.

Received August 3, 2001; accepted after revision December 12, 2001.

Address correspondence to D. G. Mitchell.

Abstract Top
Abstract
Introduction
OBJECTIVE. To determine the significance in patients with Materials and Methods
cirrhosis of small (<20 mm) hepatic nodules that show no Results
hyperintensity on T2-weighted MR images but that enhance during Discussion
References
arterial phase MR imaging, we reviewed the cases of patients with
such nodules.

MATERIALS AND METHODS. Our review of radiology reports yielded 68 nodules in 40

patients with cirrhosis that showed no hyperintensity on T2-weighted MR images but had
rapid enhancement during arterial phase MR imaging after administration of a gadolinium
contrast agent. Thirty-four patients (60 nodules) had multiple follow-up MR imaging
examinations (range of length of follow-up, 1-72 months; average length of follow-up, 15
months 2 weeks). The final diagnosis of the nodule was determined by pathology reports or
after at least 2 years of follow-up to ensure nodule stability and, therefore, benignity. Two
radiologists independently reviewed MR images of the nodules, noting the size, signal
intensity on T1- or T2-weighted images, and homogeneity of contrast enhancement.

RESULTS. Nine (13%) of the 68 nodules were hepatocellular carcinomas (HCCs). The size
of nodules on the first MR examination was between 4 and 20 mm (mean size, 9.5 mm). No
significant correlation between the diagnosis of HCC and nodule signal intensity (p = 0.48) or
contrast enhancement homogeneity (p = 0.56) on first MR examination was found. Positive
predictive value (PPV) and negative predictive value (NPV) for diagnosing HCC on the basis
of nodule growth were 100% and 98%, respectively. For diagnosing HCC on the basis of a
change in nodule signal intensity, the PPV was 60% and the NPV was 91%. For diagnosing
HCC on the basis of a change of enhancement homogeneity, the PPV was 63%, and the NPV
was 94%.
CONCLUSION. A finding of growth in small, early-enhancing nodules in patients with
cirrhosis is highly predictive of HCC. When small nodules are observed on a single
examination, close follow-up of the patient appears appropriate.

Introduction

Hepatocellular carcinoma (HCC) is the most common primary

hepatic malignancy worldwide. The incidence of HCC is higher in Top
patients with cirrhosis because of the presence of hepatitis B, Abstract
hepatitis C, or chronic alcohol abuse [1]. MR imaging plays an Introduction
Materials and Methods
important role in the detection and characterization of hepatic Results
lesions. Most cases of HCC reported in the literature have shown Discussion
high signal intensity on T2-weighted MR images and enhancement References
during the early phase of dynamic MR imaging [2,3,4,5,6,7,8,9].
However, small (<20 mm) hepatic nodules that do not show
hyperintensity on T2-weighted MR images but do enhance during arterial phase MR imaging
are often difficult to characterize. The clinical importance of these lesions is still uncertain.
The purpose of our study was to assess the clinical behavior and imaging appearance of these
small, early-enhancing nodules in cirrhotic patients.

Materials and Methods

Top
Selection Criteria Abstract
Our retrospective study was conducted according to the Introduction
institutional review board rules for departmental review of records Materials and Methods
for research. We searched records of liver MR examinations Results
performed between 1991 and 1998. The indications for imaging Discussion
References
included screening for liver lesions suspected on the basis of
findings from other imaging modalities, evaluating the severity of
cirrhosis and portal hypertension, and evaluating the patient before liver transplantation.

For inclusion in our study, a patient had to have cirrhosis that was confirmed by pathology or
by imaging criteria [10] and had to have a small (<20 mm) nodule that enhanced during the
early phase at the patient's first dynamic MR examination but did not show hyperintensity on
T2-weighted MR images (because this finding strongly indicates malignancy). Sixty-four
patients were selected using this criteria. Patients were excluded from the study if they had
undergone chemoembolization or ablation therapy (n = 6). In these patients, small nodules
might be satellite metastases. In addition, patients (n = 15) that were lost to follow-up or did
not have an adequate arterial-predominant phase (n = 3) were excluded. The criteria of
adequate arterial-predominant enhancement required for inclusion were a minimal
enhancement of the liver, no enhancement of the hepatic veins, intense enhancement of the
pancreas and renal cortex, a minimal enhancement of renal pyramids, and a serpiginous
enhancement of spleen.
Consequently, the study population included 40 patients with 68 nodules. The group included
29 men and 11 women, whose ages ranged from 31 to 78 years (mean age, 52 years), with
cirrhosis caused by viral infection (hepatitis B [n = 18], hepatitis C [n = 12], combined
hepatitis B and C [n = 1]), alcoholic hepatitis (n = 4), hemochromotosis (n = 1), autoimmune
hepatitis (n = 1), or an unknown entity (n = 3). Twenty-five patients had one nodule, nine
patients had two nodules, three patients had three nodules, one patient had four nodules, one
patient had five nodules, and one patient had seven nodules.

Six patients (eight nodules) had only one MR examination. Thirty-four patients (60 nodules)
had multiple follow-up MR examinations (range of length of follow-up, 1-72 months; average
length of follow-up, 15 months 2 weeks). The number of measurements of nodule growth in
the intervals between examinations that we had available was two measurements for 18
patients, three measurements for six patients, and four or more measurements for 10 patients.
Nine patients with 10 nodules had a follow-up MR examination less than 3 months after the
initial baseline examination, but confirmation of benign disease required a lack of growth in
or disappearance of the nodules on follow-up imaging for at least 2 years.

Diagnosis was confirmed in 14 patients by liver explantation, in 13 patients by percutaneous

biopsy, in one patient by hepatectomy, and in 20 patients by clinical means (three of these
after biopsy). Three patients underwent both liver transplantation and percutaneous biopsy.
The time interval between transplantation and final MR examination ranged from 12 to 390
days (mean, 112 days). Explanted livers were cut sequentially into 10-mm sections that
corresponded as closely as possible to the MR imaging planes. HCC and dysplastic nodules
were identified grossly as those that were distinct from the surrounding regenerative nodules
in terms of size, texture, color, or degree of bulging beyond the cut surface of the liver [11].
Clinical confirmation consisted of imaging and clinical follow-up of at least 2 years; a nodule
was considered benign if the size of the mass did not change during a sufficiently long
follow-up period [12]. Five patients underwent biopsy and had clinical confirmation of their
diagnoses.

All HCCs were pathologically confirmed by either liver transplantation (four nodules in three
patients), biopsy (four nodules in four patients), or hepatectomy (one nodule in one patient).
One patient with one nodule on MR imaging had one HCC with three dysplastic nodules
confirmed at liver transplantation.

Of the 68 nodules, 59 were proven benign. Eleven patients with 22 nodules visible on MR
images showed no corresponding lesion at explantation (of which four nodules in three
patients were also confirmed as benign at biopsy). Nine patients with 14 nodules had only
benign tissue confirmed at sonographically guided biopsy, including one patient with a
dysplastic nodule. In 20 patients with 33 benign nodules (of which six nodules in three
patients were also confirmed as benign by biopsy), follow-up imaging showed a lack of
growth in (n = 5) or disappearance of (n = 28) the nodules. Six of these 20 patients (with nine
of these 33 benign lesions) had had prior negative results at biopsy (three patients) or at
subsequent liver explantation (three patients).

MR Imaging Technique
MR imaging was performed with a 1.5-T system (Signa; General Electric Medical Systems,
Milwaukee, WI). A standard whole-body coil was used for the early examinations, but a
phased array coil was used as the receiver coil in most studies performed after 1995. All
patients underwent axial T1- and T2-weighted MR imaging. T1-weighted imaging included at
least one of the following sequences: conventional spin-echo imaging (TR range/TE, 400-
600/12) with or without fat suppression; in-phase gradient-echo imaging (80-210/4.0-4.8; flip
angle, 70-90°); and opposed-phase gradient-echo imaging (80-210/2.3-2.7; flip angle, 60-
90°). T2-weighted imaging included conventional spin-echo imaging (1500-3000/50, 90, or
100); respiratory-triggered fast spin-echo imaging (TR range/TEeff range, 3000-4000/80-102)
with fat suppression; and single-shot fast spin-echo imaging (TR/TEeff, infinite/100). The
imaging matrix used was 256 x 256, 160, or 192 pixels for the spin-echo and gradient-echo
images and 256 x 256 pixels for respiratory-triggered fast spin-echo images. A rectangular
field of view was usually used to reduce the number of phase-encoding acquisitions. The
section thickness was 7-10 mm, with an intersection gap of 1-2 mm or smaller.

Dynamic contrast-enhanced MR imaging was performed before and after administration of

gadopentetate dimeglumine (Magnevist; Berlex, Wayne, NJ). Dynamic imaging was
performed with one of two sequences—either fast multiplanar spoiled gradient-recalled echo
breath-hold imaging (70-160/2.2-2.6; flip angle, 90°) without presaturation pulse or three-
dimensional spoiled gradient-echo imaging (7-9/2-3; flip angle, 15°) with spectral fat
saturation. The contrast material dose was 0.1 mmol/kg of body weight administered as a
rapid IV bolus, followed by a 20 mL flush of normal saline. After unenhanced imaging,
arterial phase images were obtained during the 15-20 sec after the start of IV bolus
administration [12]. Second and third sets of images were obtained after allowing the patient
to take another breath, approximately 45 sec and 90 sec, respectively. The imaging matrix
was 256 x 192 pixels. The section thickness was 7-10 mm, with an intersection gap of 1-2 mm
or smaller. The time at which three-dimensional spoiled gradient-echo contrast-enhanced
imaging was initiated had been determined by a prior timing-bolus technique [13]. Three-
dimensional spoiled gradient-echo contrast-enhanced imaging was performed three times
within seconds of previous acquisition to obtain multiphasic images (arterial, early portal, and
late portal phases) for each of 18 patients. The imaging matrix was 256 x 128-160 pixels, and
the acquired slice thickness was 5-7 mm, with zero fill interpolation in the section encoding
axis to yield images at 2.5-3.5 mm increments.

Image Analysis
We reviewed the MR images, noting the size, signal intensity, and contrast enhancement
pattern of nodules. One radiologist measured the nodule size. The average size was calculated
from the greatest two dimensions in the left to right and anterior to posterior axes (measured
from axial images), as described by Havelaar et al. [14]. To test reproducibility of the data,
subsequent measurements of tumor diameter were repeated successively three times on one
patient during one of his follow-up studies.

The growth rate of HCC was expressed as tumor volume doubling time, as proposed by
Collins et al. [15], and the following formula developed by Schwartz [16] was used for
calculation: Tumor volume doubling time = time interval between measurements x log (2)/3 x
log (tumor diameter at last measurement/tumor diameter at first measurement). We estimated
the doubling time, applying the best fitting regression line derived from all of the data
available for a given patient.

Two radiologists independently evaluated the signal intensity (hyperintense, isointense, or

hypointense relative to liver parenchyma), contrast enhancement pattern (homogeneous or
inhomogeneous), and site of visible nodules. A change of signal intensity in nodules on T2-
weighted images was defined as an intensity that initially was isointense and became hypo- or
hyperintense. When disagreement in the assessments was encountered, the two radiologists
reached a consensus. The chi-square test was used to determine significant differences
between the imaging features and a diagnosis of HCC.

Results Top
Abstract
Introduction
Table 1 shows the summary of the 68 nodules at the first Materials and Methods
examination. Nine (13%) of the 68 nodules were HCCs. Eight Results
patients had nine HCCs. Thirty-four patients had 59 benign Discussion
References
lesions. Two patients had HCC and benign lesions.

View this TABLE 1 Findings of Initial MR Imaging in 40 Patients with 68

table: Small, Early-Enhancing Nodules
[in this
window]
[in a new
window]

Nodule size ranged between 4 and 20 mm (mean, 9.5 mm) at the first examination. Forty-two
nodules were smaller than 10 mm, and 26 were 10 mm or larger. Three (7%) of the 42
nodules smaller than 10 mm and six (23%) of the 26 nodules larger than 10 mm were HCC.
Fifty-two of 68 nodules were isointense on T1-weighted images. Eight of nine HCCs were
isointense on T1-weighted images. Fifty-six of 68 nodules had homogenous enhancement.
Twelve nodules had inhomogenous enhancement. Eight of nine HCCs had homogeneous
enhancement. No significant correlation existed between HCC and signal intensity (p = 0.48)
or enhancement pattern (p = 0.56).

In patients with follow-up images, eight of the 60 nodules were HCCs, and 52 were benign.
Follow-up MR findings for HCCs are summarized in Table 2. The range of length of follow-
up periods was from 2-24 months (average length, 10 months 2 weeks). On serial follow-up
MR images, the size of seven nodules increased, 24 remained stable, and 29 were no longer
visible. All nodules with interval growth were HCCs (Figs. 1A,1B,1C,1D,1E,1F and
2A,2B,2C,2D,2E,2F). One HCC, diagnosed by surgical resection, was stable for 2 months,
but all other stable nodules or nodules not detected on follow-up examinations were benign
(Fig. 3A,3B,3C,3D,3E,3F).

View this table: TABLE 2 MR Imaging Characteristics of Hepatocellular

[in this window] Carcinoma (n = 8)a
[in a new
 window]

Fig. 1A. —Hepatocellular carcinoma with interval nodule

growth in 41-year-old man. Nodule shows no change of
signal intensity or enhancement homogeneity. Axial T1-
weighted gradient-echo MR image (TR/TE, 130/4.2)
shows nodule as isointense.

View larger version

(134K):
[in this window]
[in a new window]
[as a PowerPoint slide]

Fig. 1B. —Hepatocellular carcinoma with interval nodule

growth in 41-year-old man. Nodule shows no change of
signal intensity or enhancement homogeneity. Axial fast
spin-echo MR image (TR/TEeff, 3157/80) reveals
isointense nodule.

View larger version (155K):

[in this window]
[in a new window]
[as a PowerPoint slide]
 Fig. 1C. —Hepatocellular carcinoma with interval nodule
growth in 41-year-old man. Nodule shows no change of
signal intensity or enhancement homogeneity. Axial T2-
weighted gadolinium-enhanced gradient-echo dynamic MR
image (TR/TE, 130/1.5) obtained during arterial phase of
bolus reveals early homogenous enhancement (arrow) of
13-mm nodule.

View larger version

(128K):
[in this window]
[in a new window]
[as a PowerPoint slide]

Fig. 1D. —Hepatocellular carcinoma with interval nodule

growth in 41-year-old man. Nodule shows no change of
signal intensity or enhancement homogeneity. On follow-
up axial T1-weighted gradient-echo image (120/4.2)
obtained 3 months later at the same level as C, nodule is
isointense.

View larger version

(117K):
[in this window]
[in a new window]
[as a PowerPoint slide]
 Fig. 1E. —Hepatocellular carcinoma with interval nodule
growth in 41-year-old man. Nodule shows no change of
signal intensity or enhancement homogeneity. Axial T2-
weighted fast spin-echo MR image (TR/TEeff, 7500/100)
obtained at same time as D shows nodule as isointense.

View larger version

(150K):
[in this window]
[in a new window]
[as a PowerPoint slide]

Fig. 1F. —Hepatocellular carcinoma with interval nodule

growth in 41-year-old man. Nodule shows no change of
signal intensity or enhancement homogeneity. Axial T1-
weighted gadolinium-enhanced gradient-echo MR image
(TR/TE, 120/2.1) obtained at same time as D shows
considerable growth of nodule (arrow), now measuring 21
mm.

View larger version

(111K):
[in this window]
[in a new window]
[as a PowerPoint slide]
 Fig. 2A. —42-year-old woman with hepatocellular
carcinoma. Nodule shows interval growth and change of
enhancement homogeneity. Axial T1-weighted gradient-
echo image (TR/TE, 130/2.2) shows nodule as isointense.

View larger version

(125K):
[in this window]
[in a new window]
[as a PowerPoint slide]

Fig. 2B. —42-year-old woman with hepatocellular

carcinoma. Nodule shows interval growth and change of
enhancement homogeneity. Axial T2-weighted fast spin-
echo image (TR/TEeff, 5000/100) shows nodule as
isointense.

View larger version

(127K):
[in this window]
[in a new window]
[as a PowerPoint slide]
 Fig. 2C. —42-year-old woman with hepatocellular
carcinoma. Nodule shows interval growth and change of
enhancement homogeneity. Axial gadolinium-enhanced
gradient-echo dynamic MR image (TR/TE, 7/2) obtained
during arterial phase of bolus reveals 6-mm nodule
(arrow) with homogeneous enhancement.

View larger version (114K):

[in this window]
[in a new window]
[as a PowerPoint slide]

Fig. 2D. —42-year-old woman with hepatocellular

carcinoma. Nodule shows interval growth and change
of enhancement homogeneity. Follow-up axial T1-
weighted gradient-echo MR image (TR/TE, 130/2.2)
obtained 15 months later at same level as C shows
nodule (arrow) is hypointense and now has grown to 15
mm.

View larger version (127K):

[in this window]
[in a new window]
[as a PowerPoint slide]

Fig. 2E. —42-year-old woman with hepatocellular

carcinoma. Nodule shows interval growth and change of
enhancement homogeneity. Axial T2-weighted fast spin-
echo MR image (TR/TEeff, 4000/90) obtained at same time
as D shows nodule (arrow) as more intense than liver.

View larger version

 (128K):
[in this window]
[in a new window]
[as a PowerPoint slide]

Fig. 2F. —42-year-old woman with hepatocellular

carcinoma. Nodule shows interval growth and change of
enhancement homogeneity. Axial T1-weighted gadolinium-
enhanced gradient-echo MR image (TR/TE, 7/2) obtained
at same time as D shows inhomogenous enhancement
(arrow) of nodule.

View larger version

(110K):
[in this window]
[in a new window]
[as a PowerPoint slide]

Fig. 3A. —Benign lesion in 63-year-old man that was not

seen on follow-up images. On axial T1-weighted
gradient-echo MR image (TR/TE, 120/4.2), nodule is not
visible.

View larger version (117K):

[in this window]
[in a new window]
[as a PowerPoint slide]
 Fig. 3B. —Benign lesion in 63-year-old man that was not
seen on follow-up images. On axial T2-weighted fast spin-
echo MR image (TR/TEeff, 4000/96), nodule is not visible.

View larger version

(109K):
[in this window]
[in a new window]
[as a PowerPoint slide]

Fig. 3C. —Benign lesion in 63-year-old man that was not

seen on follow-up images. Axial gadolinium-enhanced
gradient-echo dynamic MR image (TR/TE, 120/2.2)
obtained during arterial phase of bolus shows 8-mm
nodule (arrow) with homogeneous enhancement.

View larger version (140K):

[in this window]
[in a new window]
[as a PowerPoint slide]

Fig. 3D. —Benign lesion in 63-year-old man that was not

seen on follow-up images. On follow-up axial T1-
weighted gradient-echo image (150/4.2) obtained 28
months later, nodule is not visible.
 View larger version
(110K):
[in this window]
[in a new window]
[as a PowerPoint slide]

Fig. 3E. —Benign lesion in 63-year-old man that was not

seen on follow-up images. On axial T2-weighted single-
shot fast spin-echo MR image (infinite/100) obtained at
same time as D, nodule is not visible.

View larger version

(107K):
[in this window]
[in a new window]
[as a PowerPoint slide]

Fig. 3F. —Benign lesion in 63-year-old man that was not

seen on follow-up images. On axial gadolinium-enhanced
gradient-echo dynamic MR image (7/2) of left hepatic
lobe obtained at same time as D, nodule is not visible.

View larger version

(131K):
[in this window]
[in a new window]
[as a PowerPoint slide]

The doubling time in the seven HCCs with interval nodule growth had a log-normal
distribution. The mean doubling time was 2 months 2 weeks (Fig. 4). The positive predictive
value (PPV) and negative predictive value (NPV) of the interval growth of nodules for
diagnosis of HCC were 100% and 98%, respectively.

Fig. 4. —Graph shows tumor-doubling time

(longitudinal axis) in seven hepatocellular carcinomas
with interval growth versus difference of tumor
diameter in millimeters (transverse axis). Mean
doubling time was 2.47 ± 1.25 months and was less
than 5 months in all tumors.

View larger version (13K):

[in this window]
[in a new window]
[as a PowerPoint slide]

On T2-weighted images, the signal intensity of five nodules changed during the follow-up
period. Three nodules with a change in signal intensity were HCCs (Fig.
2A,2B,2C,2D,2E,2F). In the two benign nodules with a change in signal intensity, a lesion
was not identified at liver transplantation or biopsy. Fifty of 55 nodules with no change in
signal intensity were benign nodules. For diagnosis of HCC, a change in signal intensity had a
PPV of 60% and a NPV of 91%. A change of enhancement pattern was found in eight
nodules. Five of these nodules were HCCs. There was no change of enhancement pattern in
52 nodules. Forty-nine of these nodules were benign nodules. For diagnosis of HCC, a change
of enhancement pattern had a PPV of 63% and a NPV of 94%.

Discussion
Top
Abstract
HCC in the cirrhotic liver often develops in a multistep fashion, Introduction
Materials and Methods
changing from regenerating nodules through intermediate steps of Results
dysplastic nodules to HCCs [3, 11]. Early detection and diagnosis Discussion
of HCC are important because the most effective treatment for References
HCC is surgical resection or local ablation when the tumor is small
[17, 18].

Early enhancement of a nodule on arterial phase CT or MR imaging is suggestive of HCC

[6,7,8]. With improved MR imaging technique (that is, use of a timing bolus and thin
contiguous or overlapping selections), detection of tiny, enhancing foci has become more
common [5]. However, clinical treatment of patients with cirrhosis who had small nodules
detected on arterial phase imaging may be problematic. First, characterization of small
nodules is more difficult in patients with cirrhotic livers than in patients with normal livers
because of the severely disturbed liver architecture and altered portal hemodynamics caused
by cirrhosis [19]. An early-enhancing mass may also be a benign lesion such as pseudolesion,
hemangioma, or dysplastic nodule [2, 20,21,22]. The cause of early enhancement is benign
solid lesions has been suggested to be derangement in vascular flow or neoplastic
angiogenesis in the cirrhotic liver [20, 23, 24]. Second, the small size and the well-
differentiated nature of early HCC account for the frequent failure to establish definite
diagnosis using imaging or biopsy [25]. Sonographically guided biopsy is often useful in
obtaining tissue from these small nodules, but many small nodules shown on MR images may
not be visible on sonography used to perform of percutaneous biopsy, especially if the nodule
is seen only on arterial phase contrast-enhanced images. Additionally, a needle biopsy
specimen is often not sufficient for definitive histologic diagnosis because of the
multicompartmental nature of HCC and, in many cases, the similarity between the cirrhotic
liver and a well-differentiated HCC.

The finding of small, enhancing nodules in patients with cirrhosis raises questions regarding
management. Although the biological behavior of precancerous nodules associated with
cirrhosis is uncertain, they appear to have low potential for malignancy and rarely cause
metastasis or vascular invasion [26, 27]. In our study, only 13% (9/68) of small, early-
enhancing nodules were HCC. We only considered nodules that were not hyperintense to the
liver on T2-weighted images; it is more difficult to diagnose these lesions than those that are
hyperintense on T2-weighted images because the latter lesions are highly likely to be
malignant. We cannot contribute data explaining the reason that HCCs are isointense to the
liver, but we can speculate that it may be related to the histologic similarity of a well-
differentiated HCC to the liver parenchyma or to the increased signal intensity of the
surrounding damaged and inflamed liver.

It has long been hoped that MR imaging could aid in distinguishing HCC from nonmalignant
regenerative and dysplastic nodules on the basis of signal characteristics [4, 9, 28, 29].
Generally, regenerative nodules are isointense on T1-weighted images and iso- or hypointense
on T2-weighted images. Dysplastic nodules are usually hyperintense on T1-weighted images
and iso- or hypointense on T2-weighted images. Overt HCC may be hypoiso-, or
hyperintense on T1-weighted images and hyperintense on T2-weighted images, but it has
recently become clear that isointensity, or even hypointensity, of HCC on T2-weighted
images is not uncommon [4, 8, 9, 28, 30]. Kelekis et al. [31] reported that 235 (66%) of 354
HCCs were hyperintense on T2-weighted images and 106 (30%) were isointense. Thus, the
signal intensity of a dysplastic nodule and a small overt HCC overlaps considerably.

The results of previous MR imaging studies have shown the importance of imaging during
the hepatic arterial phase, especially for detection of small HCCs, because these may be
occult at other pulse sequences and on portal venous and delayed phase images [5,6,7]. Small
tumors measuring less than or equal to 1.5 cm are frequently isointense on both T1- and T2-
weighted images and may be detected only on hepatic arterial phase images as diffuse
homogeneously enhancing tumors [31]. Peterson et al. [32] reported that 16% were
prospectively detected on dynamic MR images but not on spin-echo MR images. Our series
included nine HCCs that were isointense on T2-weighted images at the first examination and
intensely enhancing on hepatic arterial phase images.

Krinsky et al. [33] studied 71 patients with cirrhosis with explantation correlation using MR
imaging. Fifteen lesions in 10 patients were incorrectly diagnosed as HCC. Four of these
lesions were seen only during the hepatic arterial phase and appeared to be well-
circumscribed masses measuring 5-21 mm. A correlating lesion could not be identified at
pathologic examination. In our study, 22 early-enhancing nodules could not be identified at
explanation. Eighteen patients had 28 nodules that were not visible on follow-up MR
imaging. Possibly, some of these early-enhancing foci represented transient vascular
phenomena such as arteriovenous shunting.

Our study showed that for diagnosis of HCC, the PPV of interval growth of nodules on
follow-up MR images was 100%. The PPVs of changed signal intensity and changed
enhancement pattern were 60% and 63%, respectively. Little is known of the growth rate of
small HCCs [34, 35] because small, early-enhancing nodules are often treated by
chemoembolization or ablation therapy when first discovered [18, 36]. Sheu et al. [34]
reported a mean doubling time of 3 months 2 weeks with HCC; a suitable screening interval
for its early detection was 4 to 5 months. In our series, all nodules with interval growth were
HCCs. The mean doubling time of HCC was 2 months 2 weeks and less than 5 months in all
cases. All but one of the nodules that were stable or not visible on follow-up images proved to
be benign. Therefore, interval serial scanning may be a practical way to eventually arrive at
the correct diagnosis of early-enhancing nodules that are not hyperintense on T2-weighted
images.

Our study had several limitations. First, because they were not visible when reexamined,
many lesions in our series were diagnosed during follow-up. Stability in the size of a solid
nodule on cross-sectional images for more than 2 years has been generally accepted as an
indicator of benignity [12]. Second, our study was retrospective, and consequently,
parameters used in the MR imaging varied. Because the T1 of liver has been recorded as
slightly less than 500 msec at 1.5 T, variations in TR, even as a TR as low as 1500 msec,
should not cause great differences of T1 contrast of subtle lesions on the T2-weighted images.
The greatest deficiency of this intermediate TR is in depicting fluid, which was not relevant in
our study. T2-weighted imaging was sometimes performed with fat suppression and
sometimes without. Nodules in cirrhotic livers may have slightly different signal intensities
with these sequences. Third, our calculations of size are based on linear measurements of the
diameter of the visualized lesions. Some error (in millimeters) is introduced because the exact
center of the small lesions might not be bisected in each scan, despite the careful use of a
measuring technique [37]. However, we were able to establish a reliable trend in tumor
growth because of the acquisition of multiple scans in some patients over an extended period.
Finally, the overall number of pathologically confirmed cases of HCC was small because the
inclusion criteria in our study were isointense nodules on T2-weighted images that showed
early increased enhancement relative to the remainder of the liver, an atypical appearance for
HCC.

In conclusion, signal intensity and contrast enhancement patterns could not be used for
definitive diagnosis of HCC in small, hepatic arterial phase—enhancing nodules in cirrhotic
patients. However, interval nodule growth is highly predictive of HCC, irrespective of the
homogeneity of enhancement. When a small, hepatic arterial phase—enhancing nodule is
observed on a single examination, close follow-up of the patient appears appropriate.