Our analysis showed that 56 of the 224 patients evaluated in this study showed extrahepatic metastases. Pulmonary metastases, regional lymphnodal, distant lymphnodal, and skeletal metastases are the most frequent sites of metastatic involvement in our population similar to as reported in literature [9]. Peritoneal/omental, splenic, muscle, and brain metastases were also seen, however, in relatively less number of patients.
18F FDG PET CT is traditionally not considered the ‘ideal’ tool for evaluation of HCC, and this is reflected in the relatively small number of consecutive patients (in this study) who were referred to us for a FDG PET CT evaluation over a period of nearly 8 years. Newer PET tracers are being investigated and a few of them such as 18F choline have become part of clinical practice. However, the synthesis of 18F choline is a relatively more expensive affair when compared to the synthesis of 18F FDG, and this is a major determinant for its regular availability especially in developing nations. It is due to this reason that most often, FDG still remains as radiopharmaceutical in use for HCC evaluation. It is in this context that this study was undertaken to demonstrate the ability of FDG PET CT in correctly identifying extrahepatic metastases in HCC. And further to use this data to highlight how this cost-effective tool is being grossly underutilized.
Pulmonary metastases were the most common of all extrahepatic metastases of HCC. Involvement of lower lobes of both lungs was the most frequent pattern as usually seen with hemogeneous metastases from extrathoracic primaries [10], possibly due to preferential blood flow induced by gravity [11]. Tumor cells are believed to transit through inferior vena cava or via lymph through main or right thoracic duct to the lungs [12]. A very small percentage of patients also had metastatic nodules in lobes of lungs other than lower lobe. Unilateral lung involvement was also seen (Table 7). Pulmonary parenchymal metastases were predominantly 1–3 cm in size. Pulmonary nodules appeared solid except in one patient where ground glass appearance was also seen. Only in 4 patients out of 56 patients (7.14%) pulmonary nodules were predominantly FDG avid with SUV maximum range of 5.56 ± 3.94 gm/ml which is fairly high enough to convincingly characterize a pulmonary lesion as a metastatic pulmonary nodule. Partial volume averaging effect, signal dilution of small lesions, restricted resolution of PET scanners, effects of respiratory motion, and low tumor cell-to-background ratio are the reasons cited for false-negative FDG uptake in metastatic pulmonary nodules [13]. In all three patients of ours where biopsy confirmed pulmonary metastases, the size of nodules was more than 3.8 cm in largest dimension. FDG positivity in equivocal small pulmonary nodules on CT can be highly suggestive of metastatic involvement as biopsy is not always feasible to confirm either due to small size or difficult locations. Pleural metastatic involvement was also seen in a few patients and is a known entity with a potential to create hemothorax [14].
Regional lymphadenopathy was seen in 44.65% of our patients. Aortocaval, paraaortic, portocaval, and left gastric stations were more commonly seen to be involved. Discrete lymphadenopathy was more common while conglomerate regional metastatic lymph nodes were also seen. Sizes of the regional nodes ranged from subcentimetric, mostly centimetric, and to maximum of 9 cm in highest dimension for conglomerate lymph nodes. Necrotic regional nodes were also seen in a small percentage of patients. Central necrosis has been suggested as a sign of malignant involvement [15]. Regional metastatic lymphadenopathy was seen in majority of our patients. However, it was interesting to see that even in absence of regional lymphnodal metastases, distant metastases were seen with or without distant metastatic lymphadenopathy (Table 11).
Distant lymphnodal involvement was seen in 39.2% of our patients which included mediastinal and even lower cervical stations. Commonest sites of distant lymphnodal involvement were paratracheal, juxtaphrenic, and mesenteric lymph nodes. Mediastinal metastatic lymphadenopathy is known to be common [15]. Other distant nodal site specific distribution of regional metastatic lymph nodes is as enumerated in Table 6. Sizes of the distant lymph nodes ranged from subcentimetric, mostly centimetric, and to maximum of 4.8 cm in highest dimension for conglomerate lymph nodes. Metastatic distant lymph nodes were predominantly discrete, and only one patient had conglomerate distant metastatic lymph nodes. Necrotic distant metastatic lymph nodes were rare, seen only in two patients. FDG avid distant lymphadenopathy was more common. However, FDG non-avid lymph nodes as well as a combination of both FDG avid and non-avid was also seen (Tables 5 and 6). The fact that distant lymphnodal metastases are predominantly metabolically active and that in HCC distant nodal/extranodal metastases are not uncommon in absence of regional metastatic lymphadenopathy, may suggest a greater role of FDG PET CT in the form of determining biopsy or FNAC site in evaluation of extrahepatic involvement of HCC. In one of our patients, a small supraclavicular lymph node of 1.5 × 1.2 cm size with an SUVmax value of 7.5 gm/ml was proven to be metastatic on FNAC.
A third of patients with HCC and extrahepatic metastasis are known to present with or develop skeletal metastasis over time [16], and they are known to be lytic, hypervascular, and often expansile with lumbosacral and thoracic spine involvement being most common [17]. Eighteen of our patients were detected with skeletal metastases (32.1%) with vertebra, pelvic bones, and sternum being the most common sites followed by ribs, scapula, and long bones (Table 8). Blood-borne tumor emboli rather than lymphatic spread is attributed for skeletal metastases and seen preferentially in bones with red marrow although coexisting skeletal metastases with neighboring nodal involvement is frequent [18]. Of the 18 patients who were detected with skeletal metastases, 8 patients had only skeletal metastases without any lymphnodal or other metastases. It is believed that increased abdominal pressure causes blood to bypass caval systems and reach vertebral plexus veins and thereafter venous and sinusoidal systems of the bones of the spine, shoulder, skull, and even elbows or knees are known to get involved. The remaining patients with skeletal metastases had lymphnodal or nodal plus other sites of involvement. All anatomically evident skeletal lesions were lytic and sometimes were associated with soft tissue component (Table 9). Sclerotic lesions were not seen in any of our patient and are known to be rare [19]. Interestingly, we had four patients in whom there was no corresponding anatomical lesion at the site of abnormally high focal metabolic activity seen on FDG PET. All skeletal lesions except one lesion were highly metabolically active. Our SUVmax range was 5.35 ± 2.11 gm/ml for skeletal lesions which is fairly high enough to convincingly characterize a lesion as metastatic especially when interpreted along with CT features. Our findings of possible isolated skeletal metastases in a small subset of HCC patients and also absence of anatomical lesions at sites of abnormal focal metabolic activity in bones red flags a possibility that conventional locoregional imaging may simply not suffice to detect skeletal metastatic lesions.
Seven patients (12.5%) were detected with adrenal gland metastases. One patient had bilateral adrenal gland involvement, a reported pattern in literature [20]. Three lesions in two patients were histologically proven. There was anatomical abnormality in seven adrenal glands in seven patients, while in the patient who had bilateral involvement, there was no anatomical abnormality in the other adrenal gland at the site of abnormal focal metabolic activity. Both these lesions were histologically proven to be metastases from HCC at different time points. Four lesions with anatomical abnormality were between 1 and 4 cm sizes whereas three lesions were above 4 cm. Two lesions showed mild enhancement on the venous phase CT, while in remaining five patients, there was no enhancement. Five patients (8.9%) had significant abnormal metabolic activity with a SUV range of 5.74 ± 2.16 gm/ml while there was no metabolic abnormality in lesions in two patients. Three lesions larger than 4 cm were all FDG avid while only two of the lesions within 1–4 cm were FDG avid. The smallest lesion which was FDG avid was 2.0 × 1.3 cm. Most of the adrenal gland metastases tended to be metabolically active even in the 1–4 cm range which exemplifies the ability of metabolic imaging to characterize small adrenal lesions.
Peritoneal deposits were detected in only two of our patients (3.5%) although a higher frequency of almost up to 16% have been reported [21]. Deposits were well defined and nodular. They were multiple in number which is a more common pattern although solitary implants are also known [21]. The highest dimensions were 1.8 cm and 2.5 cm, respectively. However, the patients did not have FDG uptake in the lesions. One of the patients had an exophytic component in the primary HCC lesion with regional metastatic lymphadenopathy. Direct invasion from an exophytic tumor and hematogenous transfer through variceal collateral pathways have both been suggested as mechanism of peritoneal dissemination [22]. The other patient was a post-liver transplant recipient and had regional metastatic lymphadenopathy and distant pulmonary, skeletal, and adrenal metastases.
Metastatic involvement of spleen in HCC is known to be rare [23]. We had only one patient with splenic metastasis. Both patterns of single and multiple discrete nodular involvement in spleen are known [18]. In our patient, splenic lesions were multiple, appearing to be hypodense and significant metabolic activity (SUVmax of 5.3 gm/ml) along with a focal metabolically active lesion in right acetabulum. These were new lesions compared to previous FDG PET along with metabolically active recurrence in the post-radiofrequency ablation (RFA) site and alpha fetoprotein(AFP)level of 10,000 ng/ml. Hematogeneous spread through splenic arterial blood flow, through the splenic vein (in patients with portal hypertension) or even by lymphatics in a retrograde fashion have been suggested [11].
Metastasis to muscles is infrequently seen in chest wall and paraveretebral muscles [18]. In one of our patients, a heterogeneously enhancing lesion measuring 5.8 cm × 2.3 cm was seen deep to the pectoralis muscles at the sternal attachment overlying the costochondal junctions with minimal erosions of the calcification along the anterior margin of 4th costochondral cartilage. There was heterogeneous FDG distribution with SUVmax of 7.4 gm/ml. New focal ill-defined hypodensities in liver were detected in comparison to previous CT and were metabolically active. The patient had presented with an increasing left anterior chest wall swelling within a period of 4 years after radiofrequency ablation of HCC in segment V/VI. Subsequently, USG-guided FNAC from anterior chest wall swelling suggested it to be a poorly differentiated neoplasm, possibly poorly differentiated carcinoma.
Central nervous system metastases are reportedly rare. Reported incidences range from 0.6 to 7.7% with brain parenchymal metastasis being the most common amongst CNS manifestations [9]. In our study, in 1 patient, a solitary enhancing lesion 9 mm × 9 mm in precentral gyrus with perilesional edema was seen post-left hepatectomy 5 years after surgery and was being investigated after presenting with convulsions. The lesion was moderately metabolically active with SUVmax of 5.0 gm/ml. This patient also had another metabolically active peripherally located solid 4.5 × 4.5 cm mass lesion in the medial segment of middle lobe of right lung with SUVmax of 9.5 gm/ml which was subsequently proven to be metastasis from HCC. There was however no local recurrence. Very rarely leptomeningeal seeding have also been reported [9].
In 26 patients who had undergone local/systemic therapy or post-liver transplant (Table 11), irrespective of the regional metastatic lymphnodal involvement status, all had distant extra nodal metastatic involvement. It is interesting to note that in this group, there were no patients with only metastatic nodal involvement without distant organ metastases. Whereas among the other 30 patients evaluated as a part of initial evaluation, there were 9 patients who only had metastatic lymphadenopathy and no distant metastases. In six patients, however, distant metastases were detected without either regional or distant nodal involvement. But there were 15 patients who had distant metastases along with either regional or distant nodal metastases or with both. The analysis of factors leading to the varied patterns of metastatic involvement was beyond the scope of our study.