Unresectable solitary hepatocellular carcinoma not amenable to radiofrequency ablation: Multicenter radiology-pathology correlation and survival of radiation segmentectomy


  • Michael Vouche,

    1. Department of Radiology, Section of Interventional Radiology and Division of Interventional Oncology, Robert H. Lurie Comprehensive Cancer Center, Northwestern University, Chicago, IL
    Search for more papers by this author
  • Ali Habib,

    1. Department of Radiology, Section of Interventional Radiology and Division of Interventional Oncology, Robert H. Lurie Comprehensive Cancer Center, Northwestern University, Chicago, IL
    Search for more papers by this author
  • Thomas J. Ward,

    1. Department of Radiology, Section of Interventional Radiology and Division of Interventional Oncology, Mount Sinai Hospital, New York City, NY
    Search for more papers by this author
  • Edward Kim,

    1. Department of Radiology, Section of Interventional Radiology and Division of Interventional Oncology, Mount Sinai Hospital, New York City, NY
    Search for more papers by this author
  • Laura Kulik,

    1. Department of Medicine, Division of Hepatology, Northwestern University, Chicago, IL
    Search for more papers by this author
  • Daniel Ganger,

    1. Department of Medicine, Division of Hepatology, Northwestern University, Chicago, IL
    Search for more papers by this author
  • Mary Mulcahy,

    1. Department of Medicine, Division of Hematology and Oncology, Robert H. Lurie Comprehensive Cancer Center, Northwestern University, Chicago, IL
    Search for more papers by this author
  • Talia Baker,

    1. Department of Surgery, Division of Transplantation, Comprehensive Transplant Center, Northwestern University, Chicago, IL
    Search for more papers by this author
  • Michael Abecassis,

    1. Department of Surgery, Division of Transplantation, Comprehensive Transplant Center, Northwestern University, Chicago, IL
    Search for more papers by this author
  • Kent T. Sato,

    1. Department of Radiology, Section of Interventional Radiology and Division of Interventional Oncology, Robert H. Lurie Comprehensive Cancer Center, Northwestern University, Chicago, IL
    Search for more papers by this author
  • Juan-Carlos Caicedo,

    1. Department of Surgery, Division of Transplantation, Comprehensive Transplant Center, Northwestern University, Chicago, IL
    Search for more papers by this author
  • Jonathan Fryer,

    1. Department of Surgery, Division of Transplantation, Comprehensive Transplant Center, Northwestern University, Chicago, IL
    Search for more papers by this author
  • Ryan Hickey,

    1. Department of Radiology, Section of Interventional Radiology and Division of Interventional Oncology, Robert H. Lurie Comprehensive Cancer Center, Northwestern University, Chicago, IL
    Search for more papers by this author
  • Elias Hohlastos,

    1. Department of Radiology, Section of Interventional Radiology and Division of Interventional Oncology, Robert H. Lurie Comprehensive Cancer Center, Northwestern University, Chicago, IL
    Search for more papers by this author
  • Robert J. Lewandowski,

    1. Department of Radiology, Section of Interventional Radiology and Division of Interventional Oncology, Robert H. Lurie Comprehensive Cancer Center, Northwestern University, Chicago, IL
    2. Department of Medicine, Division of Hematology and Oncology, Robert H. Lurie Comprehensive Cancer Center, Northwestern University, Chicago, IL
    Search for more papers by this author
  • Riad Salem

    Corresponding author
    1. Department of Radiology, Section of Interventional Radiology and Division of Interventional Oncology, Robert H. Lurie Comprehensive Cancer Center, Northwestern University, Chicago, IL
    2. Department of Medicine, Division of Hematology and Oncology, Robert H. Lurie Comprehensive Cancer Center, Northwestern University, Chicago, IL
    3. Department of Surgery, Division of Transplantation, Comprehensive Transplant Center, Northwestern University, Chicago, IL
    • Address reprint requests to: Riad Salem, M.D., M.B.A., Director, Interventional Oncology, Section of Interventional Radiology, Department of Radiology, 676 N. St. Clair, Suite 800, Chicago, IL 60611. E-mail: r-salem@northwestern.edu

    Search for more papers by this author

  • Potential conflict of interest: Dr. Kim and Dr. Salem are advisors to BTG International, and Dr. Salem has received grants from BTG. Dr. Ganger advises, is on the speakers' bureau for, and received grants from Gilead. He has received grants from Merck and Ocera. Northwestern contract research organization receives research support from BTG International for execution of randomized phase 3 trials.


Resection and radiofrequency ablation (RFA) are treatment options for hepatocellular carcinoma (HCC) <3 cm; there is interest in expanding the role of ablation to 3-5 cm. RFA is considered high-risk when the lesion is in close proximity to critical structures. Combining microcatheter technology and the localized emission properties of Y90, highly selective radioembolization is a possible alternative to RFA in such cases. We assessed the efficacy (response, radiology-pathology correlation, survival) of radiation segmentectomy in solitary HCC not amenable to RFA or resection. Patients with treatment-naïve, unresectable, solitary HCC ≤5 cm not amenable to RFA were included in this multicenter study. Administered dose, response rate, time-to-progression (modified Response Evaluation Criteria in Solid Tumors [mRECIST]), radiology-pathology correlation and long-term survival were assessed. In all, 102 patients were included in this study. mRECIST complete response (CR), partial response (PR), and stable disease (SD) were 47/99 (47%), 39/99 (39%), and 12/99 (12%), respectively. Median time-to-disease-progression was 33.1 months. In all, 33/102 (32%) patients were transplanted with a median (interquartile range [IQR]) time-to-transplantation of 6.3 months (3.6-9.7). Pathology revealed 100% and 50-99% necrosis in 17/33 (52%) and 16/33 (48%), respectively. Median overall survival was 53.4 months. Univariate analysis demonstrated a survival benefit for Eastern Cooperative Oncology Group (ECOG) 0 patients. In the multivariate model, age <65, ECOG 0, and Child-Pugh A were characteristics associated with longer survival. Conclusion: Radiation segmentectomy is an effective technique with a favorable risk profile and radiology-pathology outcomes for solitary HCC ≤5 cm. This approach may allow for treatment of HCC in difficult locations. Since RFA and resection are not options given tumor location, there appears to be a strong rationale for this technique as second choice. (Hepatology 2014;60:192–201)


American Association for the Study of the Liver


alpha fetoprotein


alanine transaminase


aspartate aminotransferase


Barcelona Clinic Liver Cancer




complete pathological necrosis


complete response


Eastern Cooperative Oncology Group


hepatocellular carcinoma


interquartile range


modified Response Evaluation Criteria in Solid Tumors


magnetic resonance imaging


overall survival


progressive disease


partial necrosis


partial response


radiofrequency ablation


stable disease


99Technetium-macroaggregated albumin


transarterial chemoembolization


United Network for Organ Sharing


90Yttrium radioembolization

Hepatocellular carcinoma (HCC) is a health problem of global proportions that has a wide variety of treatment options. Tumor location, size, number of lesions, liver function, and clinical symptoms all play a role in weighing various treatment modalities. Although liver transplantation and resection are considered curative, scarce transplant availability and patient/tumor characteristics limit those options to ∼20% of cases.[1-3] Locoregional therapies such as radiofrequency ablation (RFA), transarterial chemoembolization (TACE), and 90Yttrium radioembolization (Y90) allow accurate tumor targeting while sparing surrounding parenchyma, resulting in downstaging potential and bridging-to-transplantation strategies.[4-6] RFA is a current standard of care for lesions <2-3 cm; above 3 cm, the efficacy (local tumor control and disease-free survival) of this treatment modality is diminished.[7-9] Challenging tumor locations (hepatic dome, caudate lobe, major intrahepatic vessels, bowel, gallbladder, colon, heart) limit the feasibility of RFA.[10]

The responsiveness of HCC to radiation, while long recognized, has been limited by nontargeted tissue exposure and radiation injury.[11] With the development of advanced microcatheters and short radius of emission of Y90, solitary lesions confined to a particular liver segment may now be approached with superselective radiotherapy, permitting ablative doses to a confined liver volume.[12-15] The purpose of this study was to assess the efficacy (response, radiology-pathology correlation, overall survival) of radiation segmentectomy in solitary HCC ≤5 cm not amenable to ablation.

Materials and Methods

Patient Sample

The study was in accordance with the Health Insurance Portability and Accountability Act. Institutional Review Board approval was obtained; no funding was provided. This multicenter study consisted of patients treated by Y90 for HCC by way of radiation segmentectomy technique at Northwestern Memorial Hospital (Chicago, IL) and Mount Sinai Hospital (New York, NY) from 2005-2013. Inclusion criteria were: 1) unresectable solitary ≤5 HCC by AASLD guidelines criteria not amenable to RFA (location, proximity to critical structures) after review (and consensus) in a multidisciplinary HCC tumor board16; 2) absence of portal vein thrombosis/metastases; and 3) treatment-naïve. No formal bilirubin cutoff was applied given the highly localized and focal nature of the segmentectomy technique. Four patients had undergone a partial hepatectomy for HCC; lesions in these patients were distinct from the resection margin and presented with classical imaging findings for HCC. The analysis was performed on 102 patients (Northwestern N = 93, Mount Sinai N = 9). The dataset was closed September 30, 2013.

Y90 Radioembolization, Definition of Radiation Segmentectomy, Dosimetry

All patients were treated on an outpatient basis with 20-30 μm-sized Y90 (BTG, Canada). By previous recommendations, a planning angiogram was performed 1 week prior to treatment.[12] Vascular variants were identified, nontarget extrahepatic vessels were coil-embolized, and 99Technetium-macroaggregated albumin (99Tc-MAA) was injected for shunt detection. Radiation segmentectomy was defined as Y90 infusion limited to ≤2 hepatic segments (Couinaud).[13] Administered treatment dose (Gy) was determined on the basis of injected activity (GBq), lung shunt fraction (%), and mass of infused liver. Irradiated liver volumes were noted as areas of persistent enhancement in the treated segment(s) on T1 postgadolinium sequences (magnetic resonance imaging [MRI] scans) or postcontrast computed tomography (CT) scans on portal venous phase. Follow-up imaging was obtained at 1 month, followed subsequently by q3 months.

Specific dosage methodology for radiation segmentectomy has been previously described.[17] In brief, the concept is to apply “ablative” doses to HCC-bearing parenchyma. This is achieved by prospectively determining lobar volumes, prescribing an intended lobar dose (120-150 Gy), with ultimate injection of the activity vial(s) within the segmental feeding vessel in order to increase safety and minimize radiation to normal parenchyma. Hence, segmental doses are higher than the prescribed dose by the ratio of lobar:segmental volumes. Injection at the segmental level serves to achieve “ablative” doses and minimize nontarget perfusion. Exact determination of actual delivered dose to perfused tissue can be assessed by using the 1-month follow-up scan. Since the radiated segment(s) will exhibit radiation changes, this volume can be calculated using 3D volumetric software. The actual “ablative” dose administered can be calculated by using the accepted formula Dose = (activity infused × 50) ÷ perfused mass. More recently, with the advent of cone-beam CT, the perfused volume (and hence mass) can now be measured during mapping angiography, resulting in prospective real-time dosimetry planning for radiation segmentectomy. This methodology of using cone-beam CT has resulted in radiation segmentectomy usually being performed with 3, 5, 7, or 10 GBq vials, depending on vascular capacitance of the segmental vessel(s).

Pathology Analysis

The explant was evaluated for evidence of gross and histologic necrosis using hematoxylin and eosin stains using 0.5-1 cm slice thickness and a formal assessment of tumor viability by both centers. The treated target lesion was examined for the presence of viable neoplastic tissue by pathology. Percentage necrosis was tabulated using the following conservative classification as previously described: 1) 100%: absence of any viable tissue; 2) 50-99% necrosis: significant necrosis with presence of any viable tissue; 3) <50% necrosis: defined as minimal necrosis. Although other studies have defined >90% necrosis as complete necrosis, we classified complete pathologic necrosis only if 100% necrosis was noted by pathology. In order to report most conservatively, lesions with significant necrosis but with clusters of viable cells were classified as 50-99% necrosis.[18]

Data Analysis

Baseline clinical and tumor characteristics were assessed. Clinical adverse events following Y90 were identified. Laboratory follow-up (censored for transplantation) included liver function tests and alpha-fetoprotein. Laboratory toxicities were graded according to the Common Terminology Criteria for Adverse Events v4. Complete imaging review was conducted to assess modified Response Evaluation Criteria in Solid Tumors (mRECIST) response by two readers, with discrepancies adjudicated by consensus.[19] Local recurrence (new abnormal nodular enhancing tissue within/contiguous to areas considered as mRECIST complete response [CR]), local progression (>30% increase enhancing tissue according to mRECIST), new intrahepatic HCC (new lesions >1 cm with arterial enhancement and portal venous washout according to AASLD guidelines), extrahepatic metastases or vascular invasion were considered progression.


Baseline patient/tumor/treatment characteristics, laboratory toxicities, and clinical adverse events are presented using descriptive statistics (mean [±standard deviation]/median (interquartile range [IQR]/units (%)). Proportions of grade 3-4 laboratory toxicities were compared to baseline using the chi-square test. Time-to-progression (TTP) and time-to-local-recurrence were considered from first Y90 date. Survival curves were built using Kaplan-Meier method with log-rank test (univariate: gender, age, performance status, Child-Pugh, lesion size, radiation dose) followed by Cox-proportional regression model (P < 0.25 included) with censoring for transplantation. All statistics were conducted on MedCalc (Mariakerke, Belgium); P < 0.05 was considered significant.


Baseline and Treatment Characteristics

There were 59 males and 43 females. Median (IQR) age was 64 years (58-74). Two patients had Child-Pugh (CP) class C (score >9) with 24/102 (24%) of HCC diagnosed by biopsy. 72/102 (71%) lesions were located in the right liver lobe (segments 5, 6, 7, or 8), with most lesions not amenable to RFA being in the liver dome (segments 4, 7, 8). United Network for Organ Sharing (UNOS) stage was T1 (<2 cm) for 20 tumors and T2 (2-5 cm) for 82 tumors, with most lesions 3-5 cm in size. Eastern Cooperative Oncology Group (ECOG) was >0 in 40% (41/102), demonstrating that despite small lesions, cirrhosis patients may exhibit compromised performance status. Hence, there were technically 61 Barcelona Clinic Liver Cancer (BCLC) A and 41 BCLC C patients, the latter group due to performance status. Median (IQR) tumor size, treated volume, administered activity and dose to treatment site were 2.6 cm (2.1-3.6), 165 mL (108-240), 0.95 GBq (0.63-1.27), and 242 Gy (173-369), respectively (Table 1). An example of this technique is exhibited in Fig. 1A-C.

Table 1. Baseline Characteristics
  N = 102
Age (years)Median64
Child-Pugh classA49
Method of diagnosisImaging78
Lesion location(segment)14
Lesion size (cm)median2.6
Alpha-fetoprotein (ng/ml)≥20018
Albumin (gm/dl)<2.840
Bilirubin (mg/dl)<2.073
UNOST1<2 cm20
T22.0-3.0 cm37
3.1-5.0 cm45
Treated volume (mL)median165
Activity delivered (GBq)median0.95
Dose administered (Gy) to hepatic segment containing HCCmedian242
Figure 1.

(A) Contrast-enhanced arterial phase MRI demonstrating a surface segment 4 HCC adjacent to the gallbladder, falciform ligament, and liver capsule. (B) Angiography of segment 4 lesion where radiation segmentectomy was performed with 300 Gy. (C) Complete mRECIST tumor necrosis at 16-month follow-up. No viable tumor was found at explant.

Clinical Adverse Events and Laboratory Toxicities

In all, 53/102 (52%) patients exhibited adverse events (all grades). Among Y90-related clinical adverse events, 46 (45%) suffered fatigue, 10 (10%) pain, 8 (8%) nausea, 3 (3%) fever, 2 (2%) appetite loss, 1 (1%) dyspnea, 1 (1%) vomiting, and 1 (1%) weight loss. No major complication was observed. None required readmission. The most frequent grade 3-4 laboratory (lymphopenia, platelets, bilirubin, aspartate aminotransferase / alanine aminotransferase [AST/ALT]) toxicities from Y90 treatment until 2-year follow-up are listed in Table 2a; there were no significant toxicities that appeared during follow-up.

Three out of 13 patients with grade 3-4 bilirubin toxicity were found at baseline; one of three was transplanted within weeks and the other two improved to grade 2 toxicity. Six of 13 patients were found to have grade 3-4 bilirubin toxicity at month 3, which was an increase from three patients. However, five out of those six patients were CP B or C at baseline, suggesting preexisting poor liver function and underlying cirrhosis (Table 2b).

Table 2a. Grade 3-4 Laboratory Toxicitiesa
 Baseline1-3 MonthsP Value3-6 MonthsP Value6-12 Months12-24 MonthsPatients With ≥ 1 Abnormal Value From Baseline at Last Follow-up
  1. a

    Number of toxicity events; a single patient can have multiple events.

Absolute lymphocyte16/10218/940.8612/530.675/312/1937/102

Overall, 13 patients experienced grade 3-4 bilirubin toxicities at some time throughout the study. Three of these toxicities were present at baseline; one of these patients was transplanted and alive at 19 months while the remaining two survived for 7 (baseline Child-Pugh C) and 9 (baseline Child-Pugh B) months following Y90. Of the remaining 10 patients who continued to exhibit bilirubin toxicity at a single or multiple follow-ups, four were transplanted. While a median survival was not reached in these 10 patients, 60% were found to be alive at 30 months when censored to transplantation.

Alpha-Fetoprotein (AFP) Response

In patients with AFP >200 ng/mL, median baseline AFP was 2,333 ng/mL (range: 267-49,879); it was 469 at 1 month (range: 10-64,518; 80% reduction, P = 0.01) and 136 at 3 months (range: 2-30,000; 94% reduction; P = 0.0007). In 5 of the 18 patients (28%), AFP was completely normalized (<15 ng/mL).

Imaging Outcome

One patient was lost to follow-up and two patients had no imaging follow-up due to early transplantation. Among the 99 other patients, CR was obtained in 47/99 (47%), partial response (PR) in 39/99 (39%) and stable disease (SD) in 12/99 (12%) according to the mRECIST criteria. One progressive disease (PD) was observed at 1-month follow-up scan, demonstrating new intrahepatic lesions, vascular invasion, and numerous extrahepatic metastases. The rate and median (IQR) time-to-disease-progression was 27/102 (26%) and 33.1 months (IQR: 10-35), mostly influenced by the appearance of new intrahepatic lesions (16/27, 59%). Of those 16 patients that developed new lesions (N = 16/99 = 16%), the median time-to-new lesion was 6.2 months (2.8-12.4). Other causes of disease progression included local progression (5/27, 19%; one was associated with new intrahepatic lesions), local recurrence after mRECIST CR (5/27, 19%), and vascular invasion (2/27, 7%; one was associated with new intrahepatic lesion). No extrahepatic metastases were noted. Local recurrence occurred in 8/47 CRs. Hence, median (IQR) time-to-local-progression and time-to-local-recurrence (after mRECIST CR) were 17.1 months (4.4-27.8) and 10.5 months (1.9-15.9), respectively.

Some comments on the specifics of PD declaration in the context of RECIST and mRECIST are warranted. First, there were 16 patients who developed new nodules. They were definitively considered new HCCs when their maximal transverse diameter was ≥1.0 cm with arterial enhancement and portal venous washout. Before the finding of portal washout, those nodules with arterial enhancement only were considered dysplastic nodules; there were four such cases of new nodules presenting with arterial enhancement only. These were called dysplastic nodules in our study, while they would have been declared new tumors (as hence PD) by RECIST. The median size of the 16 newly diagnosed HCC at the time of definitive diagnosis was 1.6 cm (range: 1.0-2.4 cm). Second, local recurrence (by definition only after mRECIST CR) was declared in any new suspicious nodular enhancement adjacent to a completely treated lesion. We did not follow RECIST and wait until a >30% increase of the entire lesion size was noted since: 1) tumors in “ablated” areas usually progress within the ablated sector, and 2) waiting for the entire lesion to grow would have resulted in the purposeful observation (and hence withholding of therapy) of progressing tumor within a treated area and a misleadingly long TTP. Third, local progression (by definition only after mRECIST PR) was declared only in cases of increased of enhancing tissue by >30%.

Pathological Results

In all, 33/102 (32%) patients were transplanted with a median (IQR) time-to-transplantation of 6.3 months (3.6-9.7). Pathology findings included complete pathologic necrosis (CPN) in 17/33 (52%) and partial (50-99%) necrosis in 16/33 (48%). All of the cases of partial (50-99%) necrosis exhibited >90% necrosis by pathology assessment; this translates to radiation segmentectomy resulting in 90-100% pathology necrosis in all treated patients. More complete necrosis was observed when irradiation dose exceeded 190 Gy (P = 0.03; Table 3), suggesting the possibility of a threshold dose needed to achieve CPN. Baseline lesion size, radiation dose, mRECIST response, and pathological response of transplanted patients are summarized in Table 4.

Table 2b. Time to Occurrence of Grade 3-4 Bilirubin Toxicities
 Baseline Child-Pugh ClassTime Elapsed From Y90 (Months)
2B1 (transplanted 26 days after adverse event)
7A1 (with progressive disease)
10B10 (transplanted 69 days after adverse event)
Table 3. Pathological Outcome by Radiation Dose
  1. P = 0.03 (Fisher's exact test).

<190 Gy9312
>190 Gy71421

Of the 14 CRs that were assessed at explant, 50% resulted in CPN. This translates into the remaining 50% achieving only partial necrosis at explant, highlighting the limitations of imaging tools (in this case mRECIST) in predicting complete necrosis in the context of radioembolization.

Survival and Uni/Multivariate Analysis

At the time of data closure, 30/102 (29%) patients had died. Cause of death included tumor progression (8/30; 27%), liver failure (8/30; 27%), both progression/liver failure (1/30; 3%), variceal bleeding (1/30; 3%), indeterminate (5/30; 17%); and 7/33 (21%) patients had died after transplantation. Median overall survival was 53.4 months with a median follow-up time (reverse Kaplan-Meier) of 27.1 months (Supporting Fig. 1). Censored to transplantation, survival was 34.5 months (Supporting Fig. 2). Y90 was the only antitumoral therapy in 91% (93/102) of patients. Additional treatments included Y90 (N = 6), TACE (N = 1), and RFA (N = 3) in nine patients for local recurrent/progressive tumor. In the univariate analysis, survival benefit was observed for ECOG 0 patients. In the multivariate model, age <65, ECOG 0 and CP A were characteristics associated with longer survival (Table 5).

Table 4. Radiological-Pathological Analysis by Lesion Size and Radiation Dose
Patient NumberBaseline Lesion Size (cm)Radiation Dose to the Treatment Site (Gy)mRECIST ResponseRate of mRECIST CRPathology ResponseRate of CPN
201.3285CR2/3(66%)CPN3/4 (75%)
251.7166CR PN 
291.8790Transplanted before follow-up CPN 
331.8206PR CPN 
232.1179CR8/16(50%)PN8/16 (50%)
212.1284CR PN 
312.151PR PN 
242.1109CR PN 
112.2227PR PN 
132.3120PR CPN 
302.3131CR CPN 
322.4266CR CPN 
142.5207CR CPN 
122.5560PR CPN 
222.5219SD CPN 
62.6447PR PN 
92.6391CR PN 
82.6222PR CPN 
182.9112CR PN 
172.9260PR CPN 
153300CR3/13(23%)CPN6/13 (46%)
33184CR PN 
53256SD PN 
163202PR CPN 
263.1191CR CPN 
273.3120SD CPN 
103.397PR PN 
13.3137PR PN 
23.5227PR PN 
43.71259CR CPN 
283.8236PR PN 
73.9186SD PN 
194.6244PR CPN 
Table 5. Uni/Multivariate Survival Analyses
 Univariate AnalysesMultivariate Analyses
VariableCategoryHazard Ratio(CI)P ValueHazard Ratio(CI)P Value
Lesion size<3 cm1.000.27
3-5 cm1.59(0.69-3.68)
Radiation dose<250 Gy1.000.27
≥250 Gy0.63(0.27-1.47)

Further subgroup survival analyses were completed on the 33 patients who were transplanted. Median time-to-transplantation from treatment was 6.3 months [IQR: 3.6-9.7]. Twenty-five of the 33 patients were alive at the time of data closure. Median survival in these 33 transplanted patients was 56.5 months (95% confidence interval [CI]: 51.9, -) (4.7 years).


Tumor Control, Survival, and Bridge-to-Transplantation

The present results demonstrate the ability to achieve local control by radiation segmentectomy, supported by a high rate of imaging and pathological CR, acceptable time-to-local recurrence, and time-to-disease-progression. These findings are supported by the gold standard pathology assessment at explant analysis. Keeping in mind the limitations of studying overall survival in HCC treated by locoregional therapies (influenced by underlying liver disease, previous therapies, liver transplantation, crossover to other treatments), survival was not found to be dramatically different from RFA data (with respect to the Child-Pugh class); comparable local tumor control was obtained (confirmed by imaging and AFP) (Supporting Table).[20-22] It is of particular interest that progression was influenced more by the appearance of new intrahepatic lesions or tumor thrombus than by local progression or recurrence. The ability to maintain such a patient population (solitary lesion ≤5 cm, no vascular invasion, tumor disease limited to the liver), within transplantation criteria (UCSF, Milan) is of particular interest in a bridge-to-transplantation perspective. Given the high rate of mRECIST CR (with pathology correlation), it appears that high-dose Y90 using radiation segmentectomy may behave as an “ablative” therapy in select cases.

Nevertheless, due to differences in the nature of each therapy, response assessment after Y90 is more complex than for conventional ablative therapies. RFA and microwave ablation tumoricidal effect is mediated by hyperthermal damages with immediate or early response (coagulative necrosis). Y90, as a microembolic therapy, induces delayed response by radiation, inflammatory changes with hemorrhagic necrosis, edema, and persistent enhancement affecting the entire exposed area. Hence, underestimation of the response by imaging may have occurred in this analysis in patients with short-term imaging follow-up. A 3-month follow-up is generally considered necessary to detect a CR with Y90.

With comparable treatment efficacy profile, low toxicity, and rare adverse event rates, these results suggest the feasibility of head-to-head comparisons between conventional ablative techniques and high-dose (>190-200 Gy) radiation segmentectomy. This is of particular interest in lesions located in high-risk ablation areas. It is of note, at Mount Sinai, where treatment protocols favor RFA over arterial therapies for small lesions, that 8/9 patients exhibited dome lesions, a location where probe insertion is difficult (interposition of lower ribs, proximity of the diaphragm). Another technical advantage of radiation segmentectomy over conventional ablative therapies is the absence of percutaneous transhepatic puncture, decreasing the risk of iatrogenic tumor dissemination. In some centers, RFA is not recommended in potential transplant patients due to the theoretical risk of tract seeding.

Survival results were comparable to the ablation literature (53.4 months overall; 34.5 month censored). This suggests that “ablative” type treatments such as RFA or segmentectomy, if able to target the lesions adequately and provide an adequate margin, may impart the survival benefit expected in BCLC A patients. This novel concept suggests the possibility of “transarterial ablation” or “radiation segmentectomy,” with Y90 radioembolization being the subject of future research.

Gold standard pathology explant analysis also supported the ability of this technique to ablate tumor. For all lesion sizes and locations, only one treatment using this approach was necessary to achieve a CPN rate of 52% in lesions not amenable to RFA. When including partial necrosis, all lesions exhibited >90% necrosis. This supports the role of this technique in select patients where the recommended standard treatment of RFA cannot be applied. Finally, there was no correlation between mRECIST and pathology findings; this is consistent with our recent randomized study demonstrating the inability of imaging methodology to reliably predict CPN.[23] This is explained by the lack of a true embolic effect of Y90 and the fact that enhancement may persist despite completely necrotic tissue.[24, 25]

Adverse Events and Toxicity Profile

Adverse events (all grades) were common (53% of patients); all were mild and transient (fatigue, abdominal pain, nausea, fever). Lymphopenia and platelet toxicities were found to be the most common laboratory toxicities; many of these toxicities were present at baseline and not altered during follow-up. However, both of these laboratory values are known to be associated with the postradioembolization syndrome.[26] Y90 procedures could be performed on an outpatient basis and none required readmission. Liver function tests were not found to be dramatically altered within 2 years post-Y90. In fact, those few patients that developed severe liver insufficiency all exhibited elevated Child-Pugh score at baseline, suggestive of progression of cirrhosis to endstage liver disease per natural history. The causal relationship of segmental Y90 with resulting worsening of liver function in such a context remains to be proven.

Rationale and Comparison With Ablative Therapy

In addition to providing evidence of survival benefit, pathological review suggests better local tumor control when a dose >190-200 Gy can be achieved in the treatment area. This is of particular interest in solitary HCC limited to small liver volumes, offering the possibility of selective arterial treatment. However, some key points in treatment planning such as the familiarity with variants of normal hepatic arterial vasculature and use of cone-beam CT during the planning angiogram and treatment session appear essential in order to ensure a complete understanding of tumor perfusion (Fig. 1A,B). Although the term “segmentectomy” would imply a surgical excision of the treated segment(s), its use in this case appears fitting, as evidenced by not only excellent local tumor control with complete disappearance of the lesion by imaging, but also major atrophy of the treated segment(s) as follow-up imaging is obtained (Fig. 1C).

Technical advantages of Y90 over ablative techniques include the avoidance of transhepatic puncture (reducing the risk of tract seeding, parietal injuries, or arterio-portal fistulas) and the ability to target lesions traditionally difficult for ablation (dome, caudate lobe, hepatic hilum, gallbladder). Also, since a sector of liver tissue is perfused using this technique, microsatellites surrounding the HCC may also be treated, resulting in a treatment margin, analogous to a surgical margin (Supporting Fig. 3). This is supported by the finding of all lesions exhibiting 90-100% necrosis at pathology explant. Among potential disadvantages, exposure to ionizing irradiation (although ablation is often performed with CT guidance) and cost should be mentioned. Costs associated with Y90 are now being addressed by the concept of same-day outpatient radioembolization.[27]

While there are many studies looking at radiology-pathology in animal models, there is a paucity of data in the literature looking at ablation and ability to achieve CPN in human explants. There have been other studies reviewing pathology after ablation; these report CPN rates of 55%, 66%, and 74%.[28-30] As opposed to these three studies, our study was performed in lesions not amenable to RFA. Despite this limited ability to compare, CPN rates in this study using radiation segmentectomy in lesions not amenable to RFA were comparable to those reported in RFA studies.

Strengths and Limitations

This study faces some limitations. First, during follow-up assessment the treated volume was observed in a time-dependent manner, with enhancement from radiation dissipating with time. This possibly led to a slight overestimation of radiation dose. In addition, we also assumed uniform activity distribution in the treated volume; however, it has been shown that microsphere distribution (including with chemoembolization) is preferential to tumor when performing segmental injections.[31] Unfortunately, Y90 microsphere cannot be imaged (except by positron emission tomography [PET]), limiting possibilities of posttreatment treated volume estimation. Future radioembolization materials such as holmium-loaded spheres will likely eliminate this shortcoming.[32] Also, given the lack of macroembolic effect of Y90, imaging findings should not necessarily seek to achieve imaging CR.[24, 25] Finally, RECIST criteria were not used to document local progression since this would have necessitated the enlargement of the entire “radioablated” segment, as opposed to mRECIST, which focuses on the concept of enhancing tissue. Had we used only RECIST for progression analyses, a misleadingly prolonged TTP would have been observed.

There are strengths. First, our study reports on treatment of HCC when the recommended approach is not feasible (second line); robust data are lacking in the literature. The two centers combined have significant experience in Y90 for HCC. Inclusion criteria were very stringent, resulting in a homogenous population. Data analyses were very extensive (survival, laboratory [>2 years], imaging [until transplantation/death], dosimetry, adverse events). Gold standard pathology results were used for final determination of treatment efficacy. Given the growing trend for the use of mRECIST, we used this method with two readers and adjudication in the case of discordance; we believe the results would be similar to our previous work using the EASL concept. One advantage to the use of mRECIST we detected was to ensure that new nodules were only declared when >1 cm enhancing lesions with washout was noted; this prevented an overcall of progressive disease by labeling any new finding on imaging as a new lesion.[33] The results and lessons from this study may serve as a basis for future randomized controlled trials comparing radiation segmentectomy to ablative techniques. This is one of few radiology-pathology studies comparing imaging and pathology tissue in humans; most of the literature reports on the ability of ablation to achieve necrosis in animal models. Finally, our study confirms our previous randomized study showing the lack or correlation between imaging findings and explant tissue.[23]

In conclusion, radiation segmentectomy is a safe and efficacious technique providing excellent tumor response and CPN in solitary HCC ≤5 cm not amenable to RFA. These data support its use as a second choice if RFA or resection is not feasible. This is of particular interest in the bridge-to-transplantation perspective, where in some centers ablation is not recommended for concern of malignant tract seeding. Additional studies are required to further refine the role of radiation segmentectomy in solitary HCC. Several international randomized phase 3 studies are currently under way investigating the role of Y90 in HCC.