Efficacy and Validity of Image-Guided Percutaneous Fine Needle Aspiration and Core Biopsy of Liver Pathologies: Saga of Focal Hepatic Lesions from the Nodule to the Needle to the Slide

Abstract

Context:

Radiology and pathology are pivotal tools in the investigational artillery for management of wide spectrum of hepatic lesions and early detection is of a paramount importance.

Aims:

The study aimed at analyzing the efficacy, comparative yield and validity of image-guided aspiration cytology (FNA)/core biopsy (CB) in focal hepatic lesions.

Settings and Design:

A retrospective hospital-based study was conducted in departments of Pathology and Radiology and Imaging of a tertiary care center.

Materials and Methods:

Cases of focal hepatic lesions that underwent percutaneous image guided-FNA reported (2011-2018) were analyzed. Cytological-histopathological correlation was performed where available. FNA diagnoses were divided into four categories-positive for malignancy (group 1), atypical (group 2), negative for malignancy (group 3), and non-diagnostic (group 4).

Statistical Analysis Used:

Categorical data was depicted in the form of frequencies and proportions. Validity of percutaneous image-guided FNA diagnosis was collated with the final diagnosis and results were analyzed.

Results:

A total of 338 FNA of focal hepatic lesions were reported in which 217 (68.2%) cases in group 1; 21 (6.2%) in group 2; 58 (17.2%) in group 3 and 42 (12.4%) in group 4. CB correlation was available in 123 cases. Based on clinical, radiological and pathological findings, conclusive final diagnoses were obtained and the cases were regrouped [malignant cases-245, benign lesions-57 and uncertain lesions-36]. Metastasis was the most common malignancy (175/245; 71.4%). Sensitivity, specificity, and overall diagnostic accuracy of FNA to categorize the lesion as benign or malignant were 96.94%, 100% and 97.51%, respectively. However, the cytology-histopathology correlation revealed discordance of subtyping the lesion in 20% of cases and sensitivity and specificity reduced to 80% and 50% respectively in rendering the specific diagnosis.

Conclusions:

Percutaneous image-guided FNA is a sensitive and specific tool with high diagnostic accuracy in evaluating focal hepatic lesions. The study highlights the pre-eminence of interventional radiology and cytology in the care of patients with liver lesions.

INTRODUCTION

Liver shows an exhaustive gamut of pathology, focal and diffuse; benign and malignant; primary and metastatic. The metabolic functions of the organ and its dual vascular supply make the management of liver neoplasms a challenge. Accurate localization and characterization are pertinent for management decisions as they delineate the neoplasms that are compatible with surgical methods from those that need only palliative therapy. Radiological examination coupled with morphological assessment by fine needle aspiration (FNA) cytology and/or lesional core biopsy (CB) is the first and pivotal step in this process.[1]

Ultrasonography (USG) is exceptional in screening the hepatic lesions; however, it lacks specificity due to overlap of imaging features and is not an adequate imaging modality in the presence of background parenchymal hepatic disease such as cirrhosis.[2] Computed Tomography (CT) continues to be the workhorse of hepatic imaging and augments the recognition of the lesions with contrast enhancement (iodinated), and also helps to characterize the lesions during the imperative phase of highest distinction in attenuation between the normal hepatic parenchyma and tumour.[2] MRI is increasingly gaining acceptance in evaluating focal liver lesions. With pioneering technology like diffusion weighted imaging, apparent diffusion coefficient and magnetic resonance spectroscopy, tissue characterization can be done to arrive at a specific diagnosis. Currently liver specific contrast agents are increasingly used for precision diagnosis of various lesions.[3]

Early detection of hepatic malignancy can mitigate the dire prognosis of patients. Comprehension of hepatic perfusion is critical to differentiate hepatocellular carcinoma (HCC) from metastasis. HCC, on grounds of its hypervascularity, becomes obvious at the time of early arterial contrast sequence. On the contrary, diagnosis of metastatic conditions is conditional on the porto-venous phase for their recognition due to their hypovascularity.[4,5] In this study we scrutinized FNA and CB (wherever available) of solitary and multi-focal hepatic lesions over a period of 8 years.

The objectives of the study were: (i) to assess the efficacy of FNA and CB (wherever available) in the diagnosis of hepatic lesions, (ii) to appraise the comparative yield of FNA, and (iii) to determine the validity (sensitivity, specificity, positive predictive value, negative predictive value, and diagnostic accuracy) of percutaneous image-guided FNA in diagnosis of hepatic lesions.

MATERIALS AND METHODS

This single-center hospital-based retrospective observation review was conducted in the departments of Pathology and Radio diagnosis and Imaging. This was a time-bound study where cases diagnosed over a period of 8 years (2011-2018) were incorporated. All solitary or multiple space-occupying hepatic lesions, lesions suspicious of neoplasms or hepatoma-like lesions in cirrhosis not diagnostic on two cross-sectional imaging, that underwent either USG- guided or CT guided FNA for pathological evaluation comprised of the study sample. Histopathological correlation of cases who had undergone CB was carried out. Patients with parenchymal lesion and history of abnormal bleeding and clotting parameters, hemodynamically unstable, non-cooperative subjects, and those who were unable to breath hold and had significant ascites surrounding the liver were excluded. Institutional ethics committee consent was obtained for the study (IEC KMC MLR 09-17/169).

Patients were primed for shallow breathing and breath hold and positioned on the ultrasound couch in a comfortable lateral decubitus or supine position. A precursory planning USG of the target area was performed to determine the skin site for needle entry and trajectory to the target. The free hand technique was employed with entry point selected in a manner that the needle enters at a 45° angle into the lesion. For the FNA, a 22G spinal needle (BD Spinal Needle Quincke Type Point, Becton Dickinson India Pvt Ltd, 22G×3.50IN) was utilized. When the needle was inside the target lesion, the stylet was pulled out. FNA was performed using a 10cc suction syringe and the material was transferred to slides and one half of the slides was dropped into a jar of methanol post-haste (for Papanicolaou staining) and the other half was air dried. For the biopsy, an 18G spring loaded cutting needle automatic biopsy gun (Bard Max Core Disposable core biopsy instrument, CR Bard Inc) was availed for all cases. A customary 22 mm throw side cutting needle was used. CT-guided procedure was performed only in settings where the lesion was indiscreetly visualized on the USG [Figure 1]. Figure 1 explains the flow chart of steps implemented during radiological evaluation of hepatic lesions.[6]

Figure 1.

Flow chart of steps implemented during radiological evaluation of hepatic lesions

May-Grunwald Giemsa (MGG) stain was applied for air dried smears and Papanicolaou (Pap) stain for all the wet smears. Special stains such as Ziehl Neelsen stain for suspected tuberculosis, periodic acid Schiff stain for suspected fungal infections were performed whenever obligatory. All the cytology smears were audited and results were recorded. McHugh et al. conducted a study on 713 liver FNA specimens and in his series, based on cytological findings, the cases were categorized into 4 diagnostic groups and this categorization was found to be practical and has significant implication in clinical decision making.7 Similarly, in the present series, the cases were assigned into 4 diagnostic groups: Group 1- positive for malignancy; group 2- atypical, suspicious for malignancy; group 3- negative for malignancy and group 4- non-diagnostic. The cases where the atypical/dysplastic cells were sparse in number or where the cytomorphological features were not sufficient to be called as malignant were categorized into group 2. The cytological diagnoses were further equated with the histopathological diagnoses wherever available. Based on clinical, radiological and pathological findings (FNA/CB), final diagnosis was documented and the cases were re-grouped into either malignant, benign, uncertain or inadequate.

Statistical data was analyzed by IBM SPSS Statistics for Windows, Version 22.0. Armonk, NY: IBM Corp. Categorical data was depicted in the form of frequencies and proportions. Validity of percutaneous image-guided FNA diagnosis was collated with the final diagnosis. Sensitivity, specificity, positive predictive value, negative predictive value, and diagnostic accuracy were calculated and results were analyzed.

RESULTS

There were 338 image-guided FNAs of the liver in 8 years. The mean age of patients was 59.8 years with an age range of 5-87 years. The most common age group was 61-80 years with a total of 165 cases (48.8%). There were 223 men and 115 women with male-female ratio of 1.9:1. Altogether, 309 cases (91.4%) underwent USG-guided FNA and 29 cases (8.6%) were CT-guided FNA aspirates.

Ultrasound was performed in all the cases. Most of the lesions were hypoechoic in echotexture; some were hyperechoic. Metastases were demonstrated as multiple hypoechoic lesions and some exhibited targetoid appearance [Figure 2a-d]. Hepatic cysts were typically thin walled and anechoic. A solitary well-defined hypoechoic lesion in a background of parenchymal liver disease was the typical description of a hepatoma on USG.

Figure 2.

Ultrasonography/CT-guided findings of focal hepatic lesions: (a and b) Well-defined irregular hyperechoic lesions in both lobes – Metastases (white arrows). (c) Heterogeneous circular lesion with peripheral halo - Metastases (target-like lesion; white arrow). (d) Multiple focal lesions in right lobe and FNA needle as linear hyperechoic line targeting the lesions- Metastases. (e and f) Ultrasound of abscess showing irregular thick walled hypoechoic area. (g) Ill-defined heterogeneously enhancing hilar lesion (arrow) causing proximal biliary dilatation.- Cholangiocarcinoma (h) Non-enhancing well-defined hypodense lesion (simple cyst). (i) Portal phase-multiple discrete hypodense lesions (arrows; metastases). (j) Delayed phase-large multifocal hypodense lesions with faint capsular enhancement (arrow; multifocal hepatomas). k-n Hepatoma: Hypodense (plain scan; arrow); e) Hypervascular enhancement (arterial phase; arrows) and hypodense with faint capsular enhancement (delayed phase; arrow; g) and contrast washout (portal phase; arrow; h)

Hemangiomas were unvaryingly hyperechoic and abscesses were generally hypoechoic [Figure 2e and f]. Hemangiomas were diagnostic on CT/MRI and hence image-guided-FNA was not required for their diagnosis. Intrahepatic type of cholangiocarcinomas were located peripherally and appeared as hypo-dense masses frequently showing necrotic areas. Metastases appeared as hypodense lesions on CT which seemed more distinct in the portal and delayed venous phases. Hepatic cysts looked hypodense with no lesional enhancement. Figure 2g-j shows the CECT findings in various focal hepatic lesions. HCC had a diverse appearance on ultrasound, from hypoechoic to hyperechoic. On CT, it was noticed that HCC displayed hypervascularity, neovascularity, and arterio-venous shunting [Figure 2k-n]. An arterial hyperenhancing lesion with wash-out in the venous phases, in a background of cirrhosis was gauged diagnostic of hepatoma.

Based on only the cytological diagnosis, as described above, group 1 (positive for malignancy) had 217 (68.2%) cases; group 2 (atypical) comprised of 21 (6.2%); group 3 (negative for malignancy) consisted of 58 (17.2%) and group 4 (non-diagnostic) included 42 (12.4%) cases. Figures 3 and 4 shows the cytological features of primary hepatobiliary and metastatic hepatic lesions. Details of the cases diagnosed by FNA are discussed below. Of the 338 cases, 275 cases (81.4%) had FNA material sufficient for diagnostic interpretation either as positive or negative for malignancy and in another 21 cases (6.2%), the scant material revealed few atypical cells. In these 21 cases, seven cases had conclusive diagnosis on CB. In the rest 14 cases, only FNA report of atypical cells was available and the cases were lost to follow up.

Figure 3.

Cytological findings of primary hepatobiliary lesions: (a) Focal Nodular Hyperplasia: Benign bland-looking hepatocytes. Smears also had presence of bile-duct epithelial cells and stromal elements elsewhere (Pap; 100X). (b) Pyogenic liver abscess: Presence of neutrophils in a necrotic background (Pap; 40X). (c) Hepatocellular carcinoma: Tumour cells arranged in thick trabecular cords with transgressing capillaries (Pap; 100X). Inset: Cell block showing positive for malignant cells (H&E stain; 100X). (d) Cholangiocarcinoma: Tumour cells with pleomorphic enlarged nuclei arranged in sheets and glandular pattern (MGG;100X)

Figure 4.

Cytological findings of metastatic focal hepatic lesions: (a) Metastatic Adenocarcinoma: Glandular arrangement of pleomorphic tumour cells having hyperchromatic nuclei (Pap; 400X). (b) Metastatic Melanoma: The aspirate smears show presence of abundant melanin pigment in the background with scattered malignant cells (Pap; 40X), Inset: Large pleomorphic malignant cells with intracytoplasmic melanin in pseudo-cohesive cluster (MGG; 400X). (c) Metastatic Leiomyosarcoma: Cellular aspirate with tumour composed of fascicles of atypical spindle-shaped cells. The cells have elongated nuclei with blunt ends and uniform nuclear chromatin (Pap; 100X). (d) Metastatic Neuroendocrine carcinoma: Dyscohesive tumour cells with scant cytoplasm and granular salt and pepper nuclear chromatin (Pap; 40X). (e) Metastatic Poorly differentiated carcinoma: Sheets of undifferentiated pleomorphic tumour cells (Pap; 200X). (f) Metastatic Gastrointestinal Stromal Tumour: Dyscohesive cluster of bland to epithelioid malignant stromal cells with scant cytoplasm (Pap; 400X). (g) Metastatic IDC of Breast: Clusters of cohesive malignant cells exhibit cellular pleomorphism and nuclear atypia and hyperchromasia (Pap; 100X). Note the presence of individual cells with intact cytoplasm (Inset; Pap; 100X). (h) Non-Hodgkin Lymphoma: Monoclonal medium to large sized atypical lymphoid cells (Inset; Pap; 40X). (i) Metastatic SCC: Cellular smear with presence of polygonal cells, tadpole cells, dyskeratotic squamous cells in a background of tumour diathesis. Note the marked nuclear atypia. (Pap; 400X)

Histopathological diagnosis was existing in 123/338 cases, of which 96 cases were malignant, 9 benign, 8 uncertain and 10 were inadequate on CB. When cytological and biopsy diagnoses were equated, the FNA and CB material were adequate for interpretation in 70/123 cases (Malignant-67; benign-3). Cytology-histopathology correlation [Table 1] was performed for these 70 cases and concordance of the diagnosis was achieved in 56 cases (80%). In the rest of 14 cases, the cytological interpretation as benign or malignant was concordant with the CB diagnosis; however, the subtyping of the malignant lesion was found to be discordant. Table 1 provides the details of cytology-histopathology correlation and the diagnoses rendered. In the remaining 53 cases which had histopathological diagnosis, 34/123 cases did not have adequate FNA material (Malignant-29; benign-5), 11/123 cases had non-diagnostic CB material and 8/123 cases lacked both FNA and CB material. Of these eight cases, seven cases were clinically and radiologically suspected to have malignancy and one case was an abscess.

Table 1.

Cytology-Histopathology and Radiology-Cytology correlation

Cytology-Histopathology Correlation
Diagnosis(Total number)	Number of concordant cases	Number of discordant cases	Discordant diagnosis on FNA
Metastatic adenocarcinoma (25 cases)	20	5	HCC-1 Metastatic undifferentiated carcinoma- 2 Metastatic NEC-2
HCC (20 cases)	17	3	Metastatic adenocarcinoma-2 Metastatic Carcinoma-1
Metastatic NEC (11 cases)	8	3	Metastatic adenocarcinoma-1 Metastatic Poorly differentiated Carcinoma-2
Cholangiocarcinoma (4 cases)	2	2	HCC-1 Metastatic adenocarcinoma-1
Metastatic SCC (2 cases)	2	-	-
Metastatic Poorly differentiated Carcinoma (2 cases)	1	1	Metastatic adenocarcinoma-1
Metastatic Leiomyosarcoma (1 case)	1	-	-
Metastatic IDC (1 case)	1	-	-
Metastatic Melanoma (1 case)	1	-	-
Tuberculosis (2 cases)	2	-	-
Steatohepatitis (1 case)	1	-	-
Total number	56	14	-
Radiology-Cytology Correlation
Radiological diagnosis	Concordance on cytology	Inadequate on cytology	Discordant cases with details of final conclusion on cytology
HCC	41/91 cases	22/91 cases	28/91 cases Metastatic adenocarcinoma- 20 Abscess-1 Metastatic NEC-2 Metastatic poorly differentiated carcinoma-1 Metastatic SCC-1 Regenerative nodule-1 Cholangiocarcinoma-1
Metastasis	139/176 cases	28/176 cases	9/176 cases HCC-6 PLA-2 Cholangiocarcinoma-1
Hepatic Abscesses	9/26 cases	3/26 cases	14/26 cases Benign Hepatocytes- 5 Metastasis-4 Suspicious for malignancy- 2 HCC- 1 Cholangiocarcinoma-1 Tuberculosis-1
Cholangiocarcinoma	4/9 cases	1/9 case	4/9 cases- Metastatic adenocarcinoma
FNH	1/1 case	-	-
NHL	2/2 cases	-	-
Hydatid cyst	1/1/case	-	-

FNA - Fine needle aspiration; HCC - Hepatocellular carcinoma; NEC-Neuroendocrine carcinoma; SCC - Squamous cell carcinoma; IDC - Infiltrating ductal carcinoma; PLA - Pyogenic liver abscess; FNH - Focal nodular hyperplasia; NHL- Non Hodgkin Lymphoma

FNA based categories are discussed below with relevant histopathological correlation of each.

FNA Category 1: Positive for malignancy: A total of 217 cases were diagnosed as positive for malignancy. The frequency of wide-ranging malignant diagnoses rendered on FNA were adenocarcinoma (113,52.1%), hepatocellular carcinoma (56,25.8%), neuroendocrine tumor (22,10.1%), poorly differentiated carcinoma (9,4.1%), cholangiocarcinoma (6,2.8%), squamous cell carcinoma (5,2.3%), hematopoietic neoplasm/NHL (2,2.3%), one case each of melanoma, gastro-intestinal stromal tumor, infiltrating ductal carcinoma, and leiomyosarcoma (0.5% each).

After analyzing clinical, radiological, and pathological (FNA/CB) findings, it was noted that the most common malignancy was metastasis accounting for 150 cases. The primary sites were pancreas, stomach, lung, gall bladder, ovary, rectum, breast, colon, uterus, gastro-esophageal junction, head, and neck (pyriform fossa, larynx, maxillary sinus, and post-cricoid region), bladder, kidney, skin, prostate, and a retroperitoneal mass. Primary hepatobiliary malignancies diagnosed on FNA constituted 67 cases. Immunohistochemistry (IHC) on CB was performed on certain cases such as poorly differentiated carcinoma, hemangioendothelioma of liver, HCC, metastatic NEC, cholangiocarcinoma, metastatic SCC and also to evaluate the source of metastatic adenocarcinoma in certain cases and in a case of Non Hodgkin Lymphoma (NHL).

FNA category 2: Atypical: There were 21 cases that were interpreted as atypical on FNA. Of these, 6 cases were malignant and one case was benign on CB. Four cases had inadequate CB tissue and in another 10 cases CB was not performed. Malignant diagnoses included HCC (3 cases) and one case each of metastatic adenocarcinoma, poorly differentiated carcinoma and squamous cell carcinoma. An isolated benign case was reported to have features of alcoholic steatohepatitis.

FNA category 3: Negative for malignancy: A total of 58 FNA cases were stated as negative for malignancy. In thirty samples (51.7%), only benign hepatocytes were seen, 11 (20%) were hepatic abscesses [10-Pyogenic Liver abscess (PLA) and 1-Amoebic liver abscess], 10 (17.2%) had only inflammatory cells (8-acute and/or chronic inflammation, 2-granulomatous), 6 (10.3%) showed fibroblasts and one case showed elements of hydatid cyst. Histopathological correlation of CB was available in 16/58 cases (27.9%). There were nine cases of malignancy, four benign cases, and the remaining three were inadequate. The malignancies reported include five cases of metastatic adenocarcinoma and a case each of neuroendocrine carcinoma, poorly differentiated carcinoma, and metastatic renal cell carcinoma.

FNA category 4: Non-diagnostic: There were 42 non-diagnostic aspirates. In this category, a total of 21/42 cases (50%) had CB correlation. There were 14 malignant cases, 5 benign cases and a single case with inadequate biopsy material. The malignancies reported were metastatic adenocarcinoma (7 cases), HCC (2 cases), hemangioendothelioma (2 cases), and a single case each of SCC, cholangiocarcinoma, and NHL.

The cytology-radiology correlation is given in Table 1. A radiological diagnosis of HCC was made in 91 cases. Out of these 91 cases, 41 cases turned out to be HCC cytologically, 22 cases were non-diagnostic (few atypical or non-representative samples) and another 28 cases were discordant. Out of 28 discordant cases, 6 cases had CB diagnosis (4 cases were metastatic adenocarcinoma, 1 case of metastatic NEC and a case with inadequate material).

After excluding the uncertain and inadequate cases, the validity of FNA in diagnosing hepatic lesions was examined. It was established that hepatic FNA had a sensitivity and specificity of 96.94% and 100%, respectively in triaging the nature of lesion as benign or malignant. Positive predictive value was 100%, negative predictive value was 88.14% and hepatic FNA had an overall diagnostic accuracy of 97.51%. However, the cytology-histopathology correlation revealed discordance of subtyping the lesion in 20% of cases thus the sensitivity and specificity of hepatic FNA reduced to 80% and 50% respectively for rendering specific diagnosis under malignant category.

DISCUSSION

The epoch of superior imaging technology has resulted in increased detection of hepatic lesions. Many studies in the literature have validated that image guided FNA/CB has sensitivity of >80% and specificity ~100% in diagnosing various hepatic lesions.[7,8,9,10,11,12,13] Soudah et al. investigated 4000 cases of benign and malignant hepatic lesions and showed that FNA had 97% sensitivity.[14] FNA has been evinced to be superior compared to CB in metastatic diseases by many authors.[9,10,11,12] In the present study, sensitivity of hepatic FNA in triaging the lesion as benign or malignant was 96.94% and specificity was 100%, proving it to be reliable diagnostic tool. The results of the present series were similar to what has been documented by other authors. Malignant lesions were commoner than benign lesions and metastatic tumors were commoner than primary hepatobiliary malignancies.[7,8,9,10,11,12,13,14]

We preferred to utilize USG guidance in most of our cases, except for those lesions which were indiscrete on USG. CT guidance was used in a small percentage of cases. Hemangiomas and hepatomas, which showed typical diagnostic features on cross sectional imaging were not needled. Radiographic means of diagnosing HCC have been shown to be robustly sensitive and specific.[15]

Primary hepatic tumors are comparatively less common and liver is more commonly involved in metastasis which can be attributed to its pattern of dual vascular supply and affinity of malignant tumors metastasizing through blood.[7,16,17] Many a times, the metastatic deposits in liver may be the first to get detected and can be a lead point in investigating for the primary site of its origin. In the present study as well, metastatic tumors were commoner compared to the primary hepatobiliary malignancies. Adenocarcinoma was the most common morphological type of metastasis and pancreas was the commonest known primary site though majority of adenocarcinoma was from an unknown origin.

The utmost widely prevalent primary hepatic neoplastic malignancy is HCC.[18] Hepatic metastases might have different size and shape, and are best diagnosed in portal venous phase, when they are hypo attenuating to the liver.[19] Radiology-cytology correlation is depicted in Tables 1 and 2 shows the various radiological findings in certain focal hepatic lesions.[6]

Table 2.

Radiological and cytological findings in focal hepatic lesions

Hepatic lesions	USG findings	CT/MRI findings	Cytology findings
Hepatic Abscesses	Poorly demarcated hypoechoic lesions. Few are hyperechoic.Gas bubbles seen	Lesions are loculated, single or multiple with heterogeneous and thick rim enhancement. MRI- bright on T2W and shows restricted diffusion on DWI.	PLA-presence of neutrophils, necrotic cells and debriAmoebic Liver abscess- trophozoites of E. histolytica in a necrotic background
FNH	Homogenous iso to hypo, or hyperechoic lesions with central hyperechoic area	CT- Homogeneous, enhances strongly with hepatic arterial phase, but no washout. Central scar is iso to hypodense MRI- Hypervascular on Gadolinium, Isodense (T1), Hyperintense (T2).	Bland-looking hepatocytes with ductal-epithelial cells and stromal elements
HCC	Hypo or hyperechoic; Doppler: hypervascular	CT-Hypervascular in the arterial phase, Heterogeneous enhancement and wash out in the venous phases (hallmark) MRI- Hyper vascular, Poor differentiated, T2 hyperintense. Enhancement is similar to CT	Presence of malignant hepatocytes with presence of intracellular bile, endothelial wrapping seen
Cholangio carcinoma	Bile duct dilatation	CT- Hypodense lesion. Delayed enhancement with capsular retraction MRI- Hypointense on T1, Hyperintense on T2, MRCP is useful	Pleomorphic tumour cells in glandular pattern Nuclear overlapping and crowding
Metastasis	Hypo to hyperechoic lesions	CT- Ill-defined or well defined with subtle complete ring enhancement; more distinct on the portal phases. MRI-low intensity on T1, high intensity on T2	Dual population of hepatocytes and malignant cells Specific findings resembling primary such as dirty necrosis in colonic metastasis

Hepatic abscesses are of two types, pyogenic abscess commonly caused by E. coli through hematogenous route and amoebic abscesses through portal vein. Amoebic liver abscesses have an enhancing wall of 3-15 mm in thickness with peripheral zone of oedema giving the appearance of “double layer”.[20]

Cytological findings of the commonly encountered hepatic lesions are mostly unambiguous albeit with some overlapping features. Aspiration of normal liver parenchyma shows hepatocytes and sparse bile duct epithelial cells. Differentiating dysplastic nodule from HCC is often a diagnostic challenge in cytological smears.[21] Dominant dysplastic nodule in cirrhotic liver can resemble carcinoma on FNA with presence of atypical cells, however, if atypia is focal, the diagnosis on cytology should be deferred to biopsy with radiology correlation and classified as indeterminate. We had two cases of cirrhosis in the present series and both the cases radiologically were suspected to be hepatoma but the aspirates were inadequate. Diagnosis of cirrhosis was based on tissue biopsy. On the other end, in a background of radiologically detectable hepatic mass, a case of focal nodular hyperplasia on cytology shows the presence of benign bland-looking hepatocytes, bile-duct epithelial cells and stromal elements.[21,22]

A moderately differentiated HCC is the easiest to diagnose on cytological smears. A well differentiated HCC resembles normal hepatocytes and cell thickness with endothelial wrapping might be the pointers to look out for.[21,22,23] On the opposite end of this spectrum is poorly differentiated HCC where it is extremely difficult to identify the hepatic nature of tumor cells. To differentiate HCC from metastatic adenocarcinoma on cytology, one should look for the most differentiated cells which usually exhibit features of HCC such as intracellular bile pigment, and endothelial wrapping. Another feature which might be very helpful is a dual population of hepatocytes and malignant cells, which is seen in metastastic lesions.[21,22,23]

Cholangiocarcinoma shows specific cytological features with highly pleomorphic cells that are arranged in glandular pattern and exhibit obvious nuclear overlapping and crowding.[21,23] These features makes it easier to differentiate cholangiocarcinoma from HCC but is a diagnostic challenge to distinguish it from metastatic adenocarcinoma. Again, a well-differentiated cholangiocarcinoma exhibits very subtle pleomorphism. Therefore, clinical history, radiological findings and good clinical history aid in identifying and correlating specific morphological features.[21,22,23] Some primary tumors like colon carcinoma show clinching features like dirty necrosis. Nonetheless, ancillary tests like immunocytochemistry or immunohistochemistry form the linchpin in malignant lesions. Neuroendocrine carcinomas and poorly differentiated carcinomas irrefutably benefit from these ancillary tests by proving their differentiation as well as even pinpointing a primary site.

Figure 5 demonstrates an algorithmic approach which may be used to diagnose various focal hepatic lesions based predominantly on cytological features concurrently emphasizing the use of other tests to finalize the diagnosis. We did encounter many diagnostic conundrums, especially, towards both the ends of differentiation spectrum. Very well differentiated cholangiocarcinomas depicted such bland nuclear features, which led to a guarded cytology report. Biopsy confirmation plays a great role in these cases. These nuclear features were a recurrent problem with well-differentiated HCC as well. Thickness of trabeculae, dynamic imaging contrast features, peculiar transgressing vessels were some features that saved the day in many cases. We strongly advocate foraging scrupulously for all the features mentioned in the flowchart [Figure 5], to give a confident cytology diagnosis.

Figure 5.

Flow chart representing the algorithmic approach of diagnosing focal hepatic lesions on FNA cytology

Among the metastatic lesions, we had a low threshold for identifying the features of neuroendocrine differentiation, owing to the distinct chromatin features discerned easily in our Papanicolaou stain as well as the experience of senior cytopathologists. Mining for the primary site metastatic adenocarcinomas was, however, a challenge on cytology. Indeed, it is one of the trials even on biopsy without immunohistochemistry. Nonetheless, some worthy pointers in this regard would be dirty necrosis in colon adenocarcinomas, intracytoplasmic lumina and short cords in lobular carcinoma breast, brown black granular pigment with prominent eosinophilic nucleoli in melanoma, and clear cytoplasm in renal cell carcinoma. Attempting to predict primary site in cytology of metastatic adenocarcinoma with hepatocyte-like morphology will be most often, an inaccurate and intrepid fete, and is best deferred to ancillary tests post biopsy and clinical correlation. We might as well conclude that in no other organ is the correlation of clinical and radiology features are as much of cardinal importance as in the liver.

We attribute the cytology-histology discordance observed in this study to these diagnostic intricacies and the probable lack of complete clinical and radiological milieu at the time of cytological evaluation that usually figures early in the management strategy.

Certain rare hepatic lesions were detected in the present series such as hemangioendothelioma, melanoma, leiomyosarcoma, gastrointestinal stromal tumour, hydatid cyst and amoebic liver abscess. Many of the aforementioned tumors had concurrent CB evaluation and ancillary pathology techniques aided in confirming the diagnosis.

Many additional tests can be performed on the FNA material that may help in arriving at a definite diagnosis. The FNA aspirates can be subjected to microbiological evaluation such as culture and gene-expert in suspected cases of abscesses and tuberculosis. Cell block prepared from FNA material has many advantages such as superior cell preservation, numerous sections can be scrutinized, special stains, immunohistochemistry and FISH can be performed making it a worthy adjunct to FNA, expressly to discern primary tumours from metastasis.[24] Addition of cell block technique routinely while evaluating image-guided FNA of hepatic lesions may improve its diagnostic accuracy. Immunohistochemistry has become a quintessential tool in evaluation of focal hepatic lesions that can aid in differentiating primary from secondary tumours, HCC from cholangiocarcinoma and also to ascertain primary site of origin in metastatic cases.[25] In our series, IHC was not performed in every malignant case that underwent CB evaluation due to financial constraints. However, IHC was performed in certain challenging cases such as diagnosing hemangioendothelioma of liver, in evaluating poorly differentiated carcinoma, to differentiate between HCC and metastatic adenocarcinoma in some cases.

There were certain limitations in our study. The diagnostic accuracy of USG-guided FNA/CB, especially in poorly differentiated and doubtful tumors, can be improved by taking necessary measures such as integrating morphology with validated ancillary techniques, like immunocytochemistry which was not done in the present study. PET CT was not routinely performed and PET CT guidance was not utilized in all cases.

To conclude, the present series demonstrated the indispensable efficacy of percutaneous image-guided FNA in evaluating focal hepatic lesions with high sensitivity, specificity and diagnostic accuracy. FNA material was sufficient for diagnostic interpretation in 81.4% of cases and in another 6.2% cases atypical cells were picked up thus it had a diagnostic value in a total of 87.6% of cases. FNA yield was sufficient enough to guide further in evaluating focal hepatic lesions in 83.4% of cases. This study also highlights the importance of cytopathologists who can not only aid in rendering a definite diagnosis but also lead the light in further utilization of diagnostic material so that complete available additional information is gathered at the time of presentation for better patient management. Interventional radiologists and morphologists will continue to be the integral part of clinical management in patients presenting with focal hepatic lesions.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.