Characterization of small nodules in cirrhosis by assessment of vascularity: The problem of hypovascular hepatocellular carcinoma


  • Potential conflict of interest: Nothing to report.


In a prospective study, we examined the impact of arterial hypervascularity, as established by the European Association for the Study of the Liver (EASL) recommendations, as a criterion for characterizing small (1-3 cm) nodules in cirrhosis. A total of 72 nodules (1-2 cm, n = 41; 2.1-3 cm, n = 31) detected by ultrasonography in 59 patients with cirrhosis were included in the study. When coincidental arterial hypervascularity was detected at contrast perfusional ultrasonography and helical computed tomography, the lesion was considered to be hepatocellular carcinoma (HCC) according to EASL criteria. When one or both techniques showed negative results, ultrasound-guided biopsy was performed. In cases with negative results for malignancy or high-grade dysplasia, biopsy was repeated when an increase in size was detected at the 3-month follow-up examination. Coincidental hypervascularity was found in 44 of 72 nodules (61%; 44% of 1-2-cm nodules and 84% of 2-3-cm nodules). Fourteen nodules (19.4%) had negative results with both techniques (hypovascular nodules). Biopsy showed HCC in 5 hypovascular nodules and in 11 of 14 nodules with hypervascularity using only one technique. All nodules larger than 2 cm finally resulted to be HCC. Not satisfying the EASL imaging criteria for diagnosis were 38% of HCCs 1 to 2 cm (17% hypovascular) and 16% of those 2 to 3 cm (none hypovascular). In conclusion, the noninvasive EASL criteria for diagnosis of HCC are satisfied in only 61% of small nodules in cirrhosis; thus, biopsy frequently is required in this setting. Relying on imaging techniques in nodules of 1 to 2 cm would miss the diagnosis of HCC in up to 38% of cases. Any nodule larger than 2 cm should be regarded as highly suspicious for HCC. (HEPATOLOGY 2005.)

The definitive diagnosis of a nodular lesion, detected by imaging techniques in a liver with cirrhosis, remains a critical challenge for clinicians. The issue is particularly complicated for small (1-3 cm) nodules, because many of these may be preneoplastic lesions with uncertain malignant potential,1 such as macroregenerative nodules, low-grade dysplastic nodules (LGDN) or high-grade dysplastic nodules (HGDN), or more rarely, hemangiomas, which can be found altogether in up to 42% of cases in explanted livers.2–4 It is accepted that a number of these may contain small foci of hepatocellular carcinoma (HCC)5–7 and that most will evolve into HCC over a few years,8–10 thus making critical the decision of whether to treat these nodules.

Arterial hypervascularization, detected at contrast-enhanced imaging techniques, is now regarded as a distinctive feature of HCC in cirrhosis, because nonneoplastic nodules still have a prevalent portal vascularization.11, 12 This feature was taken into account in the European Association for the Study of the Liver (EASL) document for the clinical management of HCC published in 2001,13 where confirmation of the arterial hypervascular pattern by two imaging techniques, even in the absence of a significant (>400 ng/mL) rise in α-fetoprotein, was suggested as diagnostic criteria for HCC in nodules larger than 2 cm in patients with cirrhosis, whereas histological confirmation plays a pivotal role in the diagnosis of nodules of 1 to 2 cm because of the potential inaccuracy of imaging techniques in such small nodules. The EASL document, however, did not comment on the problem of discordant vascularity, nor on the interpretation of the hypovascular lesions. Furthermore, these recommendations have not yet been tested and validated in a prospective study aimed not only at establishing the rate of small (<3 cm) HCCs that do not satisfy the proposed noninvasive diagnostic criteria, but also at testing the impact of the more recent techniques. Further, since the establishment of these recommendations, significant improvements have been forthcoming with the advent of contrast-enhanced harmonic ultrasonography (perfusional ultrasonography [US]),14–16 and more experience has been achieved with multislice helical computed tomography (helical CT).17, 18 Magnetic resonance imaging also has evolved significantly, particularly with the use of reticuloendothelial and hepatospecific contrast agents. Therefore, it is important to establish the efficiency of the currently available advanced techniques (based on the EASL criteria) for the diagnosis of small HCC and whether the diagnostic approach to small nodules in livers with cirrhosis should be updated.

Following these considerations, we planned this prospective study, the primary aim of which was to validate the clinical impact of arterial hypervascularity as a criterion for the diagnosis of HCC in small nodules in cirrhosis detected by conventional US.

Secondary aims were: (1) to detect possible differences in the vascular pattern between nodules at first detection (incidental nodules) and those arising in patients previously submitted to curative treatment of HCC (recurrent nodules); (2) to obtain a pathological characterization of nodules appearing to be hypovascular both at perfusional US and helical CT to assess the true rate of small hypovascular HCC; and (3) to investigate the influence of the size of the nodule in the detection of arterial hypervascularity.


LGDN, low-grade dysplastic nodule; HGDN, high-grade dysplastic nodule; HCC, hepatocellular carcinoma; EASL, European Association for the Study of the Liver; US, ultrasonography; CT, computed tomography.

Patients and Methods


Recruitment started in October 2002 and continued through March 2004. All patients in Child-Turcotte-Pugh class A or B19 with a definite diagnosis of cirrhosis who were undergoing the surveillance program at 6-month intervals in our liver unit or who were referred from other institutions, as well as patients submitted to periodic controls after curative treatment for HCC, who had one to three distinct new nodules with a maximum diameter between 1 and 3 cm detected by conventional US examination were consecutively enrolled in the study. Exclusion criteria were: (1) presence of four or more nodules, (2) a nodule exceeding 3 cm in diameter, (3) local recurrence at the same site as previously treated HCC, (4) thrombosis in the main portal branches, (5) extrahepatic metastases, or (6) previous diagnosis of HCC submitted to noncurative treatment. In accordance with the EASL recommendations,13 curative treatments were considered to be surgical resection, percutaneous treatment with ethanol injection or thermoablation, and transplantation. The study population thus comprised 59 patients with a total of 72 nodules: 43 of these were detected in 38 patients without a previous diagnosis of HCC (incidental nodules), whereas 29 nodules were detected in 21 patients with other nodules of HCC previously treated with curative techniques (recurrent nodules; Table 1).

Table 1. Patient Data and Characteristics of Nodules
 Incidental NodulesRecurrent Nodules
  1. NOTE. Nodules were subgrouped as incidental or recurrent according to the liver neoplastic history. Incidental nodules are those detected in a patient without a history of focal liver lesions; recurrent nodules are those detected in patients with previously cured HCC.

  2. Abbreviations: HBV, hepatitis B virus; HCV, hepatitis C virus; PEI, percutaneous ethanol injection; RF, radiofrequency; ns, not significant.

Patient characteristics  
 Number38/59 (64%)21/59 (36%)
 Age (yr; mean ± SD)65.5 ± 9.664.8 ± 11.2
 Gender (male/female)30/817/4
 Cause (HCV/HBV/alcohol)30/3/515/3/3
 Previous treatment (PEI/RF/surgery) 4/5/12
 AFP (ng/mL; <399/>400)35/319/2
Nodule characteristics  
 Number43/72 (60%)29/72 (40%)
 Mean size (mm; ±SD)20.7 ± 517.7 ± 5 (P = n.s.)
 Pattern (hypoechoic/isoechoic/hyperechoic)26/7/1026/1/2


Conventional US examinations were performed by four experienced operators (LB, SG, NC, and FP) using a Technos scanner (Esaote, Genova, Italy). Arterial hypervascularity was assessed using perfusional US, which is superior to Doppler US in displaying the vascular pattern (Fig. 1),20 and multislice helical CT, which is the reference method of choice in this field.

Figure 1.

Typical appearance of hepatocellular carcinoma on perfusional ultrasonography (US). (A) Hypoechoic nodule (arrows) located in front of right portal branch (arrowhead) visualized on conventional US (the portal branch appears echoic because of the presence of contrast agent). Conventional US is used to target the point of interest during examination. (B) Same nodule studied simultaneously at perfusional US. In this case, at 19 seconds after contrast injection (time shown on left of image), homogeneous enhancement of the nodule (arrows) indicates hypervascularity in the arterial phase.

Perfusional US was performed using a second-generation US blood pool contrast agent (Sonovue; Bracco, Milan, Italy) and dedicated US technology (Esatune, CnTI; Esaote, Genova, Italy) with contrast-specific software operating at low acoustic pressure (mechanical index, 0.04-0.07), as described elsewhere.20 Contrast perfusion was studied during both the arterial and later phases. In each case, a diagnosis of hypervascularity was made by a consensus read by the physician performing the study and two blinded and independent readers when the nodule became hyperechoic during the early arterial phase (namely, between 10 and 45 seconds after the intravenous bolus injection of 2.5 mL of the US contrast agent) and it was distinctly detected before enhancement of the surrounding liver parenchyma (Fig. 1).

Helical CT was performed using a LightSpeed Multislice (GE Medical Systems, Milwaukee, WI) with precontrast and postcontrast triple-phase (arterial, portal venous, and equilibrium phases) helical scans, obtained after the injection of 120 mL of nonionic contrast medium (Iomeron 350 mg/I/mL; Bracco, Milan, Italy) at a rate of 3 mL/sec; scans were performed in a craniocaudal direction with a 5-mm collimation in the arterial phase and an 8-mm collimation in the other phases. Acquisition of the arterial, portal, and delayed phases was automatically started at 20, 60, and 180 seconds, respectively, after the intravenous bolus injection.

US-guided biopsy was performed using 18- or 19-gauge modified Menghini needles (Biomol; HS Laboratories, Pomezia, Italy). Specimens were routinely processed and were stained with hematoxylin and eosin and by the Masson trichrome method. Diagnosis of HCC was made according to the International Working Party criteria.1

Perfusional US examinations, CT scans, and biopsies were always performed within 1 month of US detection of the nodules.

Study Design.

The diagnostic tree adopted for this study is shown in Fig. 2. Interpretation of each of these techniques was blinded among one another. If a coincidental finding of arterial vascularization was detected by both techniques, a definite diagnosis of HCC was established according to the EASL criteria. In those cases in which one or both techniques had negative results for this finding, a US-guided biopsy of the nodule was performed (when technically possible). HCC and HGDN were submitted to treatment, the latter using percutaneous techniques. At present, HGDN diagnosed at US-guided biopsy should be regarded at least as a precursor of HCC in a number of cases.8 Macroregenerative nodules and LGDN seem to be only marginally implicated in hepatocarcinogenesis8, 10 and therefore, despite the limitations of guided biopsy, can be submitted to strict follow-up. In the latter cases, follow-up at 3-month intervals was started and biopsy was repeated whenever an increase in the size of the nodule was revealed using conventional US.

Figure 2.

Diagnostic tree for new nodules after detection on conventional ultrasonography (US). CT, computed tomography; HCC, hepatocellular carcinoma.

The study was approved by the investigation and ethics committee of our institution, with written informed consent from each patient, according to the Declaration of Helsinki.

Statistical Analysis.

Data were analyzed on a by lesion basis. Ages of patients and sizes of nodules are expressed as mean ± SD. Differences in the size of incidental and recurrent nodules were evaluated by Student t test. Differences in proportions of nodules or HCC showing various types of vascularity with imaging techniques in the various subgroups were evaluated by chi-square test. A P value less than .05 was considered significant.


Assessment of Arterial Hypervascularity in Incidental and Recurrent Nodules.

Results of imaging techniques are reported cumulatively and also separately for incidental and recurrent nodules (Table 2). Perfusional US results were positive for arterial hypervascularity in 55 of 72 nodules (76.4%), whereas helical CT results were positive in 47 of 72 nodules (65.3%). Coincidental hypervascularity with both techniques was found in 44 of 72 of all pooled nodules (61%). A contrast washout, with hypovascular appearance in later phases, was demonstrated at perfusional US in 30 of 72 nodules (41.7%; all confirmed as HCC and hypervascular in the arterial phase), whereas the remaining 25 hypervascular nodules were isovascular in portal and parenchymal phases. A similar pattern of contrast washout was also observed at helical CT in 38 of 72 nodules (52.8%; all confirmed as HCC and hypervascular in the arterial phase).

Table 2. Arterial Hypervascularity at Perfusional US and Helical CT in Incidental and Recurrent Nodules
Perfusional US/Helical CTAll Nodules (n = 72)Incidental Nodules (n = 43)Recurrent Nodules (n = 29)
  1. NOTE. No pattern of combined positivity/negativity on perfusional US and helical CT was statistically different between incidental and recurrent nodules.

  2. Abbreviations: +, hypervascular pattern detected at perfusional US (left side) or helical computed tomography (right side); −, absence of hypervascularity at perfusional US (left side) or helical CT (right side).

+/+44 (61%)28 (65%)16 (55%)
+/−11 (15%)4 (9%)7 (24%)
−/+3 (4%)2 (5%)1 (4%)
−/−14 (20%)9 (21%)5 (17%)

The separate analysis of incidental and recurrent nodules showed coincidental arterial hypervascularity in 65% of incidental and in 55% of recurrent nodules, whereas 21% of incidental and 17% of recurrent nodules were negative with both techniques (hypovascular nodules; P = not significant). α-Fetoprotein values more than 400 ng/mL were found in 5 of 59 patients (8.5%), all of whom showed nodules with an arterial hypervascular pattern both at perfusional US and helical CT.

Pathological Characterization of Nodules Without Coincidental Arterial Hypervascularization.

Of the 72 nodules, a total of 28 (39%) did not show coincidental arterial hypervascularity: 14 (20%) had negative results with both techniques (hypovascular nodules), 3 (4%) had positive results only at helical CT, and 11 (15%) had positive results only at perfusional US (Table 2).

Biopsy was technically feasible and was successfully performed in 12 of the 14 hypovascular nodules. Diagnosis of HCC was made in four nodules; the Edmonson grade21 was G1 in two nodules and G2 in two nodules. In the remaining eight nodules, the following diagnoses were established: cirrhosis (n = 2), macroregenerative nodules (n = 1), LGDN (n = 2), and HGDN (n = 3). A 6-month follow-up was achieved in four nodules, including the two that were not biopsied. One of the latter developed into HCC (G2), diagnosed on the basis of size increase and biopsy, which became feasible because of the increase in size. Biopsy was successfully performed in all 11 nodules negative for arterial hypervascularity at helical CT and positive at perfusional US; diagnosis of HCC was made in 7 nodules (1 G1 nodule and 6 G2 nodules), whereas in the remainder, the following diagnoses were established: macroregenerative nodules (n = 1), LGDN (n = 1), and HGDN (n = 2). A 6-month follow-up was reached in the two patients with macroregenerative nodules and LGDN, both of which developed into HCC (one G2 nodule and one G3 nodule) diagnosed on the basis of size increase and repeated biopsy.

Biopsy was successfully performed in two of three nodules negative for arterial hypervascularity at perfusional US and positive at helical CT; a diagnosis of HCC (G2) was established in one nodule, whereas the other was diagnosed as HGDN. A 6-month follow-up was reached in the case not biopsied that, as it increased in size, could be biopsied and developed into HCC (G2).

In summary, in the 28 nodules without coincidental hypervascularity on imaging techniques, biopsy at the time of detection led to a diagnosis of HCC in 12 nodules and follow-up in a further 4 nodules. Most nodules with only one technique with positive results for hypervascularity proved to be HCC (11/14; 78.5%), whereas among the nodules with negative results for both techniques, most proved to be nonmalignant lesions (9/14; 64%).

A final diagnosis of HCC was established in 60 of 72 nodules (44 nodules with coincidental arterial hypervascularization both at perfusional US and helical CT, 12 nodules at first biopsy and 4 at follow-up). The sensitivity of perfusional US in detecting arterial hypervascularity in nodules with a final diagnosis of HCC was not statistically different from that of helical CT (hypervascularity detected in 53/60 and in 46/60 HCC, respectively, with a sensitivity of 88% vs. 77%, respectively; P = not significant).

Influence of Size of Nodules on Detection of Arterial Hypervascularity.

Nodules detected at conventional US were divided into two different groups: those with a diameter between 1.0 and 2.0 cm (group A; n = 41) and those with a diameter between 2.1 and 3.0 cm (group B; n = 29). Findings from imaging techniques were analyzed separately in the two groups and then compared (Table 3). In group A, 29 of the 41 nodules (71%) had a final diagnosis of HCC, whereas this figure reached 100% in group B (group A vs. group B, P < .01). The rate of coincidental arterial hypervascularity, leading to a diagnosis of HCC, was significantly higher in the nodules of group B in comparison with those in group A (84% vs. 44%; P = .001). There were no hypovascular nodules with both techniques in group B, whereas these reached 34% in group A (P = .02). Analyzing only HCCs, the coincidental hypervascular pattern, with both perfusional US and helical CT, was present in 18 of 29 nodules (62%) in group A and in 26 of 31 nodules (84%) in group B (P = not significant; Fig. 3). A total of 5 of 29 HCCs (17%) in group A were hypovascular, whereas none showed this pattern in group B (Fig. 3).

Table 3. Arterial Hypervascularity at Perfusional US and Helical CT in Nodules Classified According to Size (1-2 cm vs. 2-3 cm)
 All (n = 72)Group A (1-2 cm; n = 41)Group B (2-3 cm; n = 31)P*
  • Abbreviations: +, detection of hypervascular pattern at perfusional US (perfusional US, left side) or helical CT (right side); −, absence of hypervascularity; ns, not significant.

  • *

    P values using chi-square analysis for comparison of rates of each imaging pattern in group A versus B.

Mean size (mm; ±SD) 15.5 ± 3.125.0 ± 2.4 
Perfusional US/helical CT    
 +/+44 (61%)18 (44%)26 (84%)<.01
 +/−11 (15%)7 (17%)4 (13%)ns
 −/+3 (4%)2 (5%)1 (3%)ns
 −/−14 (20%)14 (34%)0 (0%)<.01
Figure 3.

Patterns of nodules and hepatocellular carcinomas (HCCs) on cirrhosis using imaging techniques (perfusional ultrasonography [perfUS] and helical computed tomography [spCT]) according to the presence (+) or absence (−) of arterial hypervascularity.


In a relevant number of nodules, up to 17.5%,4, 22, 23 a diagnosis of malignancy is not achieved with the currently available diagnostic tools and a follow-up policy is undertaken or, alternatively, treatment is tailored individually. It is surprising how this problem, which is so common in clinical practice, has been neglected in most studies dealing with screening and surveillance.24 The correct diagnostic characterization and staging of these lesions has become even more complicated, because liver transplantation may represent the treatment of choice for most of these patients if a diagnosis of HCC is definitely established. Furthermore, in many transplant centers adopting the Model for End-Stage Liver Disease–HCC scoring system, the presence of HCC significantly increases the priority on the waiting list,25–27 thus making the precise definition of any nodular lesion mandatory.

The EASL recommendations for the management of HCC13 provide a rational approach to this problem, but leave some areas of uncertainty, particularly concerning the interpretation of discordant vascularity, the use of imaging techniques in nodules between 1 and 2 cm, the meaning of truly hypovascular nodules, and the management of nodules diagnosed as LGDN or HGDN at guided biopsy. Furthermore, the EASL criteria remain to be validated in a prospective series of patients. The present study, therefore, was aimed at clarifying the vascular pattern of small HCC using imaging techniques and validating the clinical impact of arterial hypervascularity as a criterion for the diagnosis of HCC.

For this purpose, we used the state-of-the-art technologies in the field of US and CT. In accordance with the study protocol (Fig. 2), pathological confirmation was sought and obtained in nearly all cases not fulfilling the noninvasive criteria. Whether the addition of magnetic resonance imaging (possibly with new contrast agents assessing not only the vascular pattern) may modify the diagnostic approach before biopsy will be the subject of future studies.

Vascular Pattern of Small HCC and Its Impact on Noninvasive Diagnostic Criteria.

Histopathological studies have demonstrated a progressive increase in the arterial supply to nodules evolving toward malignancy,28–30 providing the clues for the clinical diagnosis with imaging techniques.20, 31–35 To give a solid basis for the noninvasive diagnosis, the EASL panel of experts determined that this pattern must be confirmed by two different imaging techniques in nodules larger than 2 cm. No reference was made to vascularity for nodules between 1 and 2 cm, and it was pointed out that these lesions require histopathological investigation. Hence, the EASL criterion of vascularity seems to be applicable only for lesions larger than 2 cm.

Because this represents an important issue that was left open to individual interpretation, we decided to analyze the impact of arterial vascularity also in the group of nodules between 1 and 2 cm, and we were able to confirm that biopsy is required in most of these cases. Whether coincidental hypervascularity in these small lesions or even in nodules larger than 2 cm always correspond to HCC36 was not addressed in this study and remains open to further investigation.

Data obtained from perfusional US and helical CT showed a coincidental arterial hypervascularity, satisfying the EASL noninvasive criteria, in 61% (44/72) of small (<3 cm) nodules in cirrhosis and in 73% (44/60) of small nodules with a final diagnosis of HCC, demonstrating that guided biopsy is frequently required in this setting. This is particularly true for nodules of 1 to 2 cm, where fulfillment of the criteria drops to 44%, making the need for biopsy quantifiable in more than half of the 1- to 2-cm nodules at first observation. Biopsy demonstrated HCC in 11 of 23 of these cases (48%). This high rate may still be an underestimation, because HCC was detected during a short-term (3-6 months) follow-up in a few other cases. Furthermore, we are aware that the gold standard for assessing diagnostic accuracy is the meticulous examination of the explanted liver, which might have detected foci of HCC in those nodules diagnosed as HGDN. However, this gold standard is not applicable for the purpose of the present study, aimed at validating the EASL noninvasive diagnostic criteria in current clinical practice, in which most patients are not candidate for transplantation and HGDN were immediately treated.

Because some studies have claimed that the analysis of later vascular phases with contrast US37 may add a significant contribution to the characterization of HCC, we also analyzed these patterns. Our results confirm that a contrast washout leading to hypoattenuation in later phases is the most frequent and typical pattern of HCC with both techniques.

As far as hypovascular nodules at both techniques are concerned, 9 of 14 nodules (64%) proved to be nonmalignant lesions, whereas 5 nodules (all with a diameter between 1.5 and 2 cm; 36%), were truly hypovascular HCC. Hence, hypovascular HCC represents 8.3% of nodules in our series of small (<3 cm) HCCs (5/60), and 17% (5/29) in the subgroup of HCCs with a diameter of 1 to 2 cm. Hypovascular HCC were shown by other authors to be less invasive than the hypervascular type.38

In some nodules (14/72; 19.4%), the vascular pattern at perfusional US was different from that at helical CT (with more nodules being positive only at perfusional US). This discrepancy may be because of the different vascular distribution of the US and CT contrast agents, the capacity of perfusional US, by working in continuous real-time, to detect hypervascularity lasting for only a short time or occurring very early, or both. Perfusional US, however, can alsodisplay intranodular portal flow coming from the superior mesenteric vein in the following few seconds, thus making a false-positive result of early hypervascularity, and it has a limitation in depicting vascularity in deeply located small nodules. Apart from these considerations, the detectability of vascularity in small HCC was nearly identical between perfusional US and CT, as also reported by others.39

The problem of the recall policy for nodules arising in patients known to have had HCC in the past (recurrent nodules) is another issue which was not discussed in the EASL recommendations. Since this represents an important clinical challenge, due to the high number of non neoplastic nodules coexisting with HCC, as shown in explanted livers,4 we also stratified nodules into incidental and recurrent and analyzed their vascular pattern separately which proved to be similar. This finding is consistent with the theory that the majority of new nodules arising in livers with cirrhosis and previous HCC are not metastases from the primary lesion, but rather new lesions and it implies that the diagnostic approach must be the same.

The Problem of Nodules Without Confirmation of Malignancy at US-Guided Biopsy.

In a number of cases of hypovascularity shown with both techniques or without coincidental arterial hypervascularity, a diagnosis of HCC was, nonetheless, confirmed, whereas others did not prove malignant even with US-guided biopsy. The latter cases, however, which are not so rare in clinical practice, remain an unsolved diagnostic challenge. Should the finding of hypervascularity with only one technique be regarded as false-positive results? This is difficult to accept if we consider that, in the present investigation, HCC was not confirmed at first biopsy in only 5 of 14 nodules with arterial hypervascularity using only one technique, and that 3 of these 5 were found to be HGDN and the remaining 2 nodules developed into HCC during the 6-month follow-up. When making a clinical decision concerning the management of these cases, we should also take into account the lack of knowledge about the early stages of the natural history of HCC and the limitations and pitfalls of guided biopsy, which may lead to a number of false-negative results as high as 10%.40 Small foci of HCC may be present in at least one third of dysplastic nodules,6, 41 and these can be missed by biopsy. Different grades of dysplastic changes could also affect different parts of the nodule. Little is known about the possibility of correctly making a definite differential diagnosis of LGDN, HGDN, and HCC from a small biopsy specimen sampled from a part of a nodule.9 To confirm this difficulty further, even on surgical specimens, a significant interobserver variability of interpretation among pathologists was recently reported.42

In our series, we decided to treat nodules diagnosed as HGDN at guided biopsy. We are aware that not all clinicians agree with this policy. However, from a clinical and ethical standpoint, we considered it debatable whether simply to leave these patients in a follow-up program, when curative techniques that carrying a low risk of complications, such as percutaneous ethanol ablation and thermoablation, are available. This policy was also proposed in a previous paper.43

The problem remains open for nodules of 1 to 2 cm that appear to be truly hypovascular (34% in our series), because in 5 of 14 (36%), a diagnosis of HCC nonetheless was established, thus corresponding to a rate of hypovascular HCC of 17% in nodules of 1 to 2 cm with cirrhosis in our series (5 of 29 nodules). To summarize, if a clinician were to exclude malignancy in a 1- to 2-cm liver nodule in a patient with cirrhosis merely on the basis of two completely negative contrast imaging findings, he or she would carry a risk of a false-negative diagnosis of more than 35%. This risk is even greater if we consider the possibility of false-negative results at biopsy. At present, however, there is no definite answer to whether a much earlier diagnosis would imply a better outcome, but again, in the perspective of liver transplantation, earlier diagnosis would probably lead to a better chance of being placed on the waiting list and being transplanted successfully.

In conclusion, results of this study show that a number of truly hypovascular HCCs do exist, as previously shown,43, 44 particularly in nodules of 1 to 2 cm, and a not insignificant number of nonmalignant hypovascular lesions may also be found in livers with cirrhosis.4 These lesions, together with those nodules that appear to be hypervascular with only one technique, without a pathological diagnosis of malignancy, represent an area of clinical uncertainty that accounts for at least 39% of newly diagnosed small nodules in cirrhosis. These uncertainties cannot be easily solved within the EASL recommendations, especially if a biopsy does not yield a definite diagnosis of HCC. New diagnostic parameters based on different physical and pathological alterations, such as those provided by new technologies and nonvascular agents of magnetic resonance imaging, also should be considered in the diagnostic work-up of small HCC.4, 45–48 At present, an aggressive policy should, in our opinion, be considered if arterial hypervascularization with one technique is detected and other benign hypervascular lesions, such as hemangioma, angiomyolipoma, and focal nodular hyperplasia, are reasonably ruled out. Similarly, any new nodule in a liver with cirrhosis with a diameter of more than 2 cm should always be regarded as very suspicious for HCC, because the benign nature of nodules of this size (a situation not encountered in our series) is rare.49