Left-liver hypertrophy after therapeutic right-liver radioembolization is substantial but less than after portal vein embolization


  • Potential conflict of interest: Dr. Garlipp consults for and received grants from Sirtex. Dr. Pech consults for and received grants from Sirtex. Dr. Max Seidensticker received grants from Sirtex. Dr. Ricarda Seidensticker received grants from Sirtex. Dr. Ricke received grants from Sirtex. Dr. Amthauer consults for and received grants from Sirtex. Dr. van Buskirk consults for Sirtex.


In patients with liver malignancies potentially amenable to curative extended right hepatectomy but insufficient size of the future liver remnant (FLR), portal vein embolization (PVE) of the tumor-bearing liver is used to induce contralateral liver hypertrophy but leaves the tumor untreated. Radioembolization (RE) treats the tumor in the embolized lobe along with contralateral hypertrophy induction. We performed a matched-pair analysis to compare the capacity for hypertrophy induction of these two modalities. Patients with right-hepatic secondary liver malignancies with no or negligible left-hepatic tumor involvement who were treated by right-lobar PVE (n = 141) or RE (n = 35) at two centers were matched for criteria known to influence liver regeneration following PVE: 1) baseline FLR/Total liver volume ratio (<25 versus ≥25%); 2) prior platinum-containing systemic chemotherapy; 3) embolization of segments 5-8 versus 4-8; and 4) baseline platelet count (<200 versus ≥200 Gpt/L).The primary endpoint was relative change in FLR volume from baseline to follow-up. Twenty-six matched pairs were identified. FLR volume increase from baseline to follow-up (median 33 [24-56] days after PVE or 46 [27-79] days after RE) was significant in both groups but PVE produced significantly more FLR hypertrophy than RE (61.5 versus 29%, P < 0.001). Time between treatment and follow-up was not correlated with the degree of contralateral hypertrophy achieved in both groups. Although group differences in patient history and treatment setting were present and some bias cannot be excluded, this was minimized by the matched-pair design, as remaining group differences after matching were found to have no significant influence on contralateral hypertrophy development. Conclusion: PVE induces significantly more contralateral hypertrophy than RE with therapeutic (nonlobectomy) doses. However, contralateral hypertrophy induced by RE is substantial and RE minimizes the risk of tumor progression in the treated lobe, possibly making it a suitable modality for selected patients. (Hepatology 2014;59:1864–1873)


analysis of covariance


analysis of variance


future liver remnant


magnetic resonance imaging


portal vein embolization




selective internal radiation therapy

Surgical resection remains the only option for cure in patients with primary or secondary malignant liver tumors. Unfortunately, insufficient size of the liver parenchyma that can be preserved (at least 20%-25% in otherwise healthy and 40% in prediseased livers) may preclude surgery.[1] For these cases, preoperative portal vein embolization (PVE) of the tumor-bearing liver lobe has been established over the last 25 years as the standard procedure to induce hypertrophy of the nonembolized liver, resulting in an increase in future liver remnant (FLR) volume of up to 69%.[2] It has been demonstrated that a high degree of post-PVE FLR hypertrophy is associated with a decreased rate of postoperative liver dysfunction[3] and that preoperative PVE has a beneficial effect on postoperative morbidity in patients with chronic liver disease undergoing right hepatectomy.[4] However, PVE carries a significant risk of postinterventional tumor progression and may even stimulate tumor growth in the embolized as well as in the nonembolized liver lobe.[5-7] This results in 6.4%-33% of patients becoming unresectable due to tumor progress after PVE.[8, 9]

Selective internal radiation therapy (SIRT) or radioembolization (RE) uses the principle of delivering high doses of radiation to liver tumors through infusion of 90Y-labeled glass or resin microspheres into the hepatic arterial circulation of the tumor-bearing liver, thus taking advantage of the preferentially arterial blood supply of the tumor tissue. The technique has been demonstrated to yield high response rates in patients with primary and secondary liver tumors[10, 11] and is currently being evaluated for use in the first-line setting in patients with metastatic colorectal cancer in two randomized-controlled trials.[12, 13]

Besides the tumoricidal effect, there have been reports on hepatic volume changes induced by unilobar RE, notably significant hypertrophy of the contralateral, nontreated lobe in recent years.[14-19] It has been hypothesized that unilateral RE may achieve a similar extent of contralateral hypertrophy as PVE,[14] in which case it might even be preferable to PVE since it reduces the risk for tumor progression in the treated lobe; however, these modalities have never been directly compared. We therefore analyzed the degree of contralateral liver hypertrophy induction in two cohorts of patients matched for factors known to have an impact on the hepatic hypertrophy response after PVE who underwent either right-lobar PVE or right-lobar RE.

Patients and Methods

Study Design

This retrospective matched pair analysis used pooled data from two centers (center 1: University Hospital Magdeburg, center 2: Institut de cancérologie Gustave Roussy). Patients from center 1 were treated by RE, patients from center 2 by PVE. This study was approved by the local Ethics Committees; the informed consent requirement was waived because of the study's retrospective design. The primary endpoint was the relative increase in volume of the nonembolized liver after RE or PVE (referred to as the FLR in both groups even though RE was performed in a palliative setting and patients receiving RE were not planned to undergo surgery). Secondary endpoints were absolute FLR volume change and absolute as well as relative change in FLR ratio, calculated according to the equation previously published by de Baere et al.20:

display math

Furthermore, toxicities and tumor response after RE were documented.

Patient Characteristics, Eligibility, and Match Criteria

For the RE group from center 1 all patients treated by RE between October 2006 and March 2012 (n = 320) were screened. Inclusion into the analysis was based on the following criteria: 1) no underlying liver cirrhosis; 2) secondary liver malignancy, no primary liver malignancy; 3) magnetic resonance imaging (MRI) of adequate quality before and ∼4-6 weeks after RE; 4) disease limited to liver segments IV-VIII or minimal additional disease in liver segments II+III (≤2% tumor load); 5) no liver targeted therapy 3 months prior to RE or between RE and follow-up imaging; 6) availability of liver-specific laboratory parameters and clinical data before and after RE. Thirty-five patients met these criteria. The PVE population comprised 141 patients. The characteristics and clinical outcome of this patient group has been previously published by de Baere et al.[20] Briefly, this patient cohort consisted of noncirrhosis patients with secondary liver malignancies who were deemed to require hypertrophy induction in preparation for extended right hepatectomy due to insufficient volume of the FLR. Out of the identified RE and PVE patients, matched pairs were generated according to the following predefined match criteria: i) baseline FLR ratio (<25 versus ≥25%); ii) history of platinum-containing chemotherapy (yes versus no); iii) platelet count (<200 versus ≥200 Gpt/L); iv) degree of embolization (embolization of segments V-VIII versus embolization of segments IV-VIII). Only full matches were accepted. A total of 26 pairs could be matched by these criteria. Patient characteristics of the matched patients are displayed in Table 1.

Table 1. Patient Characteristics
Baseline Characteristics
Variablen or mean (median)SDn or mean (median)SDP Value
  1. a

    Comparing “one” vs. “more than one.”

  2. b

    Taxol, epirubicine, anthracyclin, cyclophosphamide, gemcitabine, mytomycin c, vinorelbine, etoposide, trastuzumab.

  3. c

    Stomach, anal, laryngeal, urothel, prostate, oropharyngeal, gallbladder.

FLR ratio at baseline (< 25/≥ 25%)14/12 14/12 1.000
Platelet count at baseline (< 200/≥ 200 Gpt/l)9/17 9/17 1.000
Prior platinum containing chemotherapy (yes/no)17/9 17/9 1.000
Segment IV embolized (yes/no)6/20 6/20 1.000
FLR at baseline (mL)368.7 (339)142.2381.7 (323)166.00.763
FLR ratio at baseline24.0 (24.4)7.925.5 (24.1)6.80.480
Liver volume embolized at baseline (mL)1157 (1153)259.91101 (1083)264.20.460
Total liver volume at baseline (mL)1525 (1447)308.31489 (1446)373.80.712
Ratio embolized to total liver volume at baseline0.76 (0.76)0.080.74 (0.76)0.070.511
Induction time (days)46.4 (46)10.234.8 (33)7.7<0.001
Platelet count at baseline (Gpt/l)272.9 (257)133.5206.7 (211)630.026
Sex (f/m)14/2 9/17 0.264
Age (years)59.2 (58)11.156.1 (56)7.10.235
Age >65y (yes/no)9/17 3/23 0.097
Weight (kg)76.7 (79)13.474.5 (73)14.60.601
Height (cm)168.7 (169)8.7171.5 (173)7.90.267
Applied Y90-activity (GBq), RE group only1.2 (1.2 range: 0.8-1.7)0.25n.a.n.a.n.a.
Chemotherapy pre treatment (yes/no)26/0 26/0 1.000
Prior chemotherapy lines (n)    <0.001a
One6 24  
Two12 2  
Three3 0  
Four4 0  
> four1 0  
Prior chemotherapy agents (n)     
5-FU/FA15 23  
Oxaliplatin10 14  
Cisplatin4 3  
Carboplatin3 0  
Irinotecan9 3  
Capecitabine3 0  
Bevacizumab6 0  
Cetuximab7 0  
Primary cancer (mCRC/other)12/14 22/4 0.008
Primary cancer exact (n)     
CRC12 22  
Breast8 2  
Othersc6 2  

Technique of 90Y-RE

RE was performed employing Yttrium-90 resin microspheres (SIR-Spheres, Sirtex Medical, Lane Cove, Australia). 90Y is characterized by a mean energy of 0.96 MeV and a half-life of 64 hours. It is coupled to resin microspheres (20 to 60 μm) and infused selectively by way of the hepatic arteries using a transfemoral approach. Treatment including preprocedural diagnostic work-up was performed according to a standard algorithm (detailed description[21]). The activity of 90Y resin microspheres was calculated by the body surface area (BSA) method. For all patients activity was first calculated as for whole-liver treatment and then adjusted for right unilobar treatment according to volume and tumor involvement of the right lobe.[22] 90Y resin microspheres were delivered selectively into the right hepatic artery. All patients received proton pump inhibitors (pantoprazole, 20 mg daily), low-dose prednisolone (5 mg daily), and ursodeoxycholic acid (500 mg daily) for 8 weeks to attenuate the effect of possibly migrated spheres in the gastric mucosa and the embolization effect to the liver parenchyma.[23]

Technique of PVE

The technique of PVE is described by de Baere et al.[20] In brief, access to the portal system was obtained with a left segmental portal branch puncture under sonographic guidance and the right segmental portal veins were selectively catheterized and subsequently embolized one after the other using n-butyl cyanoacrylate suspended in iodized oil. All the branches feeding the tumor-bearing liver were targeted for embolization. Complete occlusion of the targeted portal branches was assessed at the end of the embolization procedure with a direct portography obtained with the tip of the catheter placed in the portal trunk.

No patient underwent any type of chemotherapy in the interval between RE/PVE and post-RE/PVE liver volumetric evaluation.

Image Assessments and Volumetry

RE Group

Routine baseline and follow-up imaging consisted of MRI (1.5T, Achieva, Philips, Best, The Netherlands) using the hepatocyte selective contrast agent Gd-EOB-DTPA (Primovist, Bayer Healthcare, Leverkusen, Germany) prior to (median 16 days before RE, range 1-29 days) and ∼6 weeks (median 46 days, range 27-79 days) after RE.

For liver and tumor volumetry (whole liver, FLR, right lobe, tumor for each lobe) a 3D-T1-weighted gradient echo (GRE) sequence (slice thickness 3 mm) obtained 20 minutes after intravenous administration of 0.025 mmol/KG Gd-EOB-DTPA was used. Volumetry and metric measurements were performed with the image processing software Osirix (©Antoine Rosset, 2003-2011).

PVE Group

A computed tomography (CT) scan was performed within a week before PVE and within 3 days before planned liver surgery (median 33 days, range 24-56 days after PVE). After injection of 100 mL iodinated contrast media, maximum 5-mm slices were acquired during the portal phase from the entire liver. The total liver (excluding tumor) and FLR (excluding tumor) were delineated on each image, volumes were calculated automatically by a workstation (Advantage; GE Medical Systems, Milwaukee, WI), taking into account the delineated liver surfaces and slice thickness.

RE-Specific Analyses

Clinical and Toxicity Assessments at Baseline and During Follow-up

All patients had undergone standard clinical and laboratory examinations including liver-related parameters at first presentation and during follow-up after RE. Total serum bilirubin was analyzed before and after embolization. Additionally, the Common Terminology Criteria for Adverse Events (CTCAE) v. 4.02 (National Cancer Institute, USA) were used for toxicity assessments of laboratory values and clinical findings.

Tumor Response After RE

The tumor response after right-liver RE was evaluated using RECIST 1.1 at the time of volumetric measurements (baseline and follow-up 6 weeks after RE). Additionally, the course of disease in the FLR was evaluated according to RECIST 1.1.

Statistical Analyses

Summary statistics of baseline (pretreatment) continuous variables are provided by treatment group and overall as mean ± SD. Frequency distributions of baseline categorical variables are presented by treatment group and overall. P-values for treatment comparisons were obtained using a one-way analysis of variance (ANOVA) for continuous variables, Fisher's Exact test for dichotomous categorical variables, and chi-square general association test for nominal categorical variables.

Analysis of the association between baseline covariates and changes (and percent changes) in FLR volume and FLR ratio used an analysis of covariance (ANCOVA) model that includes baseline FLR ratio as the covariate.

Pre- and posttreatment summary statistics are presented with their change (and percent change), along with least-squares (LS) means (and medians) and their standard errors.


Out of the RE and PVE cohorts of 35 and 141 patients, respectively, 26 fully matched pairs according to the predefined criteria could be generated. Thus, the results of all further comparative analyses are based on the matched cohort of 52 patients.

Patient characteristics according to the treatment groups are displayed in Table 1. Significant differences between the groups were seen in the interval between treatment and follow-up imaging (median [range]; RE, 46 [27-79] days; PVE, 33 [24-56] days, P < 0.001), prevalence of colorectal cancer as the primary cancer site (RE, 12/26 patients; PVE, 22/26 patients, P = 0.008), chemotherapy history (more patients with more than one line of prior chemotherapy in the RE group [20/26] compared to the PVE group [2/26]; P < 0.001), and platelet count (mean; RE, 272.9 GPt/L, PVE, 206.7 GPt/L, P = 0.026). The results of covariate testing of baseline variables (including the match criteria) for possible influence on the primary endpoint variable (relative change in FLR volume after treatment) are displayed in Table 2. With the exception of treatment (RE or PVE), no factor with a significant independent impact was found in the final model. In addition, no significant correlation between the time from treatment to follow-up and the degree of contralateral hypertrophy achieved was found in either group (P = 0.351 and P = 0.135 for the PVE and RE groups, respectively).

Table 2. Association of Baseline Characteristics and FLR % Change Post-RE or PVE
VariableP ValueaP Valueb
  1. a

    Univariate analysis of categorical (ANCOVA) and continuous (linear regression) baseline variables, association of baseline variables and FLR percent change.

  2. b

    Multivariate ANCOVA modeling testing independent impact of baseline covariates on FLR percent change.

Match criteria:  
FLR ratio at baseline (< vs. ≥ 25 %)0.814 
Platelet count at baseline (< vs. ≥ 200 Gpt/l)0.240 
Prior platinum containing chemotherapy (yes/no)0.769 
Segment IV embolized (yes/no)0.300 
Treatment (RE vs. PVE)<0.0010.022
Age (continuous)0.623 
Weight (continuous)0.435 
Height (continuous)0.0230.141
primary cancer (mCRC vs. other)0.0360.640
Induction time (< vs. ≥ 40 days)0.0360.583
Induction time (continuous)0.473 
Prior chemotherapeutic lines (≤ vs. >1)0.0490.560
Prior chemotherapeutic lines (continuous)0.236 
Platelets at baseline (continuous)0.075 

In Table 3 the absolute and relative changes in FLR volume in both groups are summarized. In the RE as well as in the PVE group a significant increase in FLR volume from baseline was noted; however, the relative increase in FLR volume was significantly more pronounced in the PVE group compared to the RE patients (61.5% versus 29%, P < 0.001). Similar results were obtained for the FLR ratio, which also significantly increased from baseline in both groups but to a greater extent in the PVE group compared to the RE group (52% versus 30%, P < 0.001, data not shown).

Table 3. Group Comparison: Absolute and Relative Change of FLR After Treatment
VariableMean (median)SDMedian (median)SDP Value
FLR baseline (mL)368.7 (339)142.2381.7 (323)166.00.763
FLR post treatment (mL)470.6 (435)203.6589.5 (535)221.9 
Change from baseline (mL)101.9 (80)106.5207.9 (176)114.7<0.001
Change from baseline (%)29 (25.3)22.961.5 (50.6)37.3<0.001
P value (change from baseline within treatment, both mL and %)<0.001 <0.001  

When subjects were analyzed according to four categories of relative FLR increase after treatment (≤10%/>10% and ≤20%/>20% and ≤30%/>30%), more patients in the PVE group achieved an FLR increase >30% than in the RE group (PVE; 22 patients, RE; 12 patients). Distribution of patients in the other categories were as follows: ≤10%: 0 versus 5 patients; >10% and ≤20%: 1 versus 6 patients; >20% and ≤30%: 3 versus 3 patients in the PVE and RE group, respectively.

In the full analysis set of all patients irrespective of matching the mean (SD)/median relative FLR changes in the PVE and RE cohorts were 69 (45.5)/57% and 28.4 (20.8)/28.4%, respectively, with a P-value for change from baseline within each treatment and between treatments of <0.001 and <0.001, respectively.

In both groups a significant decrease in volume of the embolized liver was noted (RE; −125 mL, P = 0.002; PVE; −138 mL, P < 0.001) with no significant difference between treatments. The total liver volume showed no significant change after either treatment. The ratio of embolized to total liver volume (at baseline not different between the groups) showed a significant decrease after treatment in both groups. The decrease was significantly more prominent after PVE compared to RE (P < 0.001) (Table 4).

Table 4. Changes of Embolized and Total Liver Volume After RE and PVE
 REPVEP Valuea
VariableMean (median)SDMean (median)SD 
  1. a

    RE vs. PVE.

Liver volume embolized at baseline (mL)1157 (1153)259.91101 (1083)264.20.460a
Liver volume embolized at follow-up after RE/PVE (mL)1061 (1028)305966 (945)250.80.294a
P value (baseline to follow-up)0.002 <0.001  
Total liver volume at baseline (mL)1525 (1447)308.31489 (1446)373.80.712a
Total liver volume at follow-up after RE/PVE (mL)1531 (1558)3541566 1432)3910.214a
P value (baseline to follow-up)0.847 0.059  
Ratio embolized to total liver volume at baseline0.76 (0.76)0.080.74 (0.76)0.070.511a
Ratio embolized to total liver volume at follow-up after RE/PVE0.69 (0.68)0.10.62 (0.64)0.09<0.001a
P value (baseline to follow-up)<0.001 <0.001  

Laboratory Values and Toxicities After RE

No significant difference was noted between baseline and follow-up bilirubin levels in the RE patients (baseline, 7.9 [4.1-27.2] μmol/L; follow-up, 8.3 [4.2-20.8] μmol/L; P = n.s., upper limit of the norm 21 µmol/L).

Two grade 3 toxicities were recorded: leukopenia (n = 1) and acute cholecystitis (n = 1). The leukopenia built up on the basis of a preexisting chemotherapy-induced bone marrow toxicity and resolved under treatment with G-CSF. The cholecystitis developed 1 day after RE due to a significant embedding of Y-90 spheres in the cystic artery. After percutaneous drainage of the gall bladder, the situation resolved completely.[24]

Response After Right-Liver RE

According to RECIST 1.1, stable disease in the embolized lobe was seen at follow-up imaging in 19 of 26 RE patients. Partial response was seen in five patients, complete response or progressive disease each occurred in one patient.

Of the 18 patients with preexisting minimal tumor load in the FLR, one patient demonstrated new lesions in the FLR at follow-up. Preexisting lesions in the FLR had increased in size in 10 and remained stable in seven patients at follow-up. One patient without tumor involvement in the FLR at baseline developed tumor lesions in the FLR at follow-up.


Along with studies consistently demonstrating high tumor response rates in patients treated with Y-90-labeled microspheres, several reports on volume changes in treated and nontreated areas of the liver have been published in recent years.[14, 15, 17-19, 25] All of these authors reported some degree of increase in volume of the contralateral liver lobe after unilateral RE, leading to a marked interest in further evaluation of a possible therapeutic use of this phenomenon in a surgical context.[14, 15] Until now, PVE has been the standard tool to increase FLR volume prior to extensive hepatic resections and therefore we felt that a formal comparison of RE and PVE regarding their respective capacities for hypertrophy induction in the nontreated liver was warranted. To our knowledge, no such study has been published to date.

Development of contralateral hypertrophy following PVE is influenced by several patient- and treatment-related factors, making it impossible to perform a retrospective head-to-head comparison of two patient cohorts treated with either RE or PVE at two different centers. Bearing in mind that a prospective, randomized trial would be optimal in evaluating this issue but would require a large patient cohort and several years to complete, we considered matching patients according to the factors known to influence the degree of hypertrophy following PVE and performed a retrospective analysis, an appropriate method to generate valuable information to allow for the most effective development of future research strategies.

Across publications, it has been consistently demonstrated that the most pronounced liver volume gain after PVE is achieved in patients with the smallest baseline FLR ratio.[20, 26, 27] This may be due to the fact that the greater the proportion of the liver that is embolized, the more pronounced is the alteration in liver perfusion, leading to a greater extent of cytokine release and redistribution of portal venous blood flow. Second, several authors agree that platinum-containing chemotherapy prior to PVE has a negative impact on regenerative capacity (and thus, hypertrophy induction) in the liver while results with nonplatinum containing chemotherapies as well as targeted therapies have been controversial.[28-33] Platinum-containing chemotherapy is known to induce sinusoidal obstruction syndrome, which may alter portal venous blood flow by itself and thus interfere with the effect induced by subsequent PVE.[20, 34, 35] Also, embolization of segment IV together with the right-lobar hepatic segments may influence contralateral volume response, although this could not be reproduced in all trials.[20, 36-38] Finally, platelet count has been shown to be predictive of the hepatic regenerative capacity and hypertrophy induction following PVE in experimental and clinical studies, possibly due to its influence on the availability of several platelet-derived growth factors.[39] Taking into account these published data, the criteria for matching patients in the present study were chosen. Other factors known to interfere with regenerative capacity of the liver (e.g., liver cirrhosis, portal hypertension, severe hyperbilirubinemia) were not taken into account since patients with these features were excluded a priori from the analysis.

This study demonstrated that, in a cohort of 52 individuals fully matched according to these criteria, unilateral RE resulted in a mean FLR volume increase of 29%, while the mean FLR volume gain induced by PVE was 61.5%. The obvious conclusion from these figures is that PVE remains the standard treatment if maximum FLR volume gain is the goal. However, the hypertrophy-inducing effect of RE should not be disregarded and may be of use in specific patients. The most relevant aspect is that, to date, RE (in contrast to PVE) is primarily used to achieve tumor response, not to induce contralateral hypertrophy. Patients receiving RE are not typically surgical candidates. In our study as well as in most other reports, RE was used as salvage therapy in heavily pretreated patients in a palliative setting using standard therapeutic doses of Y-90, the goal being to treat the tumor without compromising liver function of the normal parenchyma in the treated lobe. With this approach, RE of the contralateral lobe remains an option for a later stage should tumor progression occur there. Thus, induction of hypertrophy was observed as a “side effect” of a treatment that was primarily designed to achieve tumor response and prevent tumor progression in the embolized lobe. This is in line with most published studies evaluating liver volume changes after RE, as standard therapeutic radiation doses were used in almost all of them. Notably, substantial hypertrophy of the nonembolized lobe but no correlation between the applied dose and the extent of hypertrophy obtained were demonstrated in these studies.[14, 16, 17, 19, 25] Therefore, we believe that if employed in patients with a small FLR who are potential candidates for surgery, this treatment may open up a curative option for some of these patients and still preserve all further palliative treatment options if cure cannot be achieved. It is likely (and has been demonstrated[18, 40]) that with substantially higher cumulative activities of 90Y applied in multiple treatment sessions, the contralateral volume response will be even more pronounced as more periportal fibrosis is induced and portal venous blood is deviated to the contralateral lobe to a greater extent; however, this comes at the cost of ablating functional liver parenchyma (“radiation lobectomy”) which may hamper further treatment if contralateral or extrahepatic progression occurs and the setting remains palliative. Moreover, since there is currently no validated method of predicting if and to what extent contralateral hypertrophy will develop after unilateral RE, deliberately applying a dose that will virtually ablate the functional right liver lobe together with the tumor may only be considered if an FLR of sufficient size to compensate for the loss of hepatic function is present from the start,[41] and therefore this approach is not feasible in patients who are considered for hypertrophy induction due to a small-sized FLR since fatal postinterventional liver failure may result. It is important to understand that even the FLR volume gain of roughly 30% observed with RE in our study as well as in other reports,[16] while being clearly inferior to the PVE results, may still convert many patients to resectability. In the full analysis set of RE patients entered into our study (n = 35), 9 of the 18 individuals who had a baseline FLR ratio <25% had an FLR ratio >25% at follow-up, indicating that volume gain induced by RE may be sufficient to achieve resectability in a substantial proportion of patients. Given the fact that PVE has been demonstrated to increase the tumor growth rate in the treated lobe,[7] RE as a means of hypertrophy induction may be preferable to PVE in patients whose lesions are at risk of becoming unresectable due to invasion of hilar structures or the left hepatic vein if tumor progression occurs, whereas in patients who need maximum FLR volume increase and whose potential plane of resection is not immediately threatened, PVE may remain the treatment of choice (Fig. 1).

Figure 1.

Suggested treatment pathway for patients requiring FLR hypertrophy induction prior to liver surgery.

There are several limitations to our study which are due to its retrospective nature. The cohorts compared were treated in very different settings. While the majority in the PVE cohort had metastatic colorectal cancer potentially amenable to curative surgery, the RE cohort consisted of heavily pretreated patients with a variety of secondary liver tumors treated with palliative intent. However, there are no data pointing toward an impact of the kind of the liver tumor upon volume response after PVE and publications on the association of prior chemotherapy with liver growth following PVE have yielded varying results. Duration of previously administered chemotherapy (≤1 versus >1 line) had no independent impact as a covariate on hypertrophy induction in our study (Table 2) and patients were matched for history of platinum-containing chemotherapy in accordance with published data, thus minimizing any bias possibly inflicted by different patient history. Because RE is used as a salvage treatment in most centers, an optimal cohort of RE patients to compare with the patients treated with PVE in our study is not available to date. It may also be criticized that the contralateral hypertrophy results obtained in the PVE cohort in our study are in the uppermost range of published figures and other studies have reported substantially less contralateral hypertrophy after PVE.[2] It is likely that this reflects the fact that our PVE patients were treated at a highly specialized center having a very long experience with this treatment and, therefore, the comparison between results obtained with unilateral RE or PVE might have been different if the PVE cohort had been treated at a less experienced institution. However, it is our opinion that a technology that is new in the field should always be tested against the existing standard performed under optimal conditions in order to compare the treatment effect itself and not just different levels of process quality. Because the interest in contralateral hypertrophy induction through RE is relatively new and data on this phenomenon are still sparse and have been obtained in very heterogeneous treatment settings,[14, 16, 17, 19, 25] it is difficult to specify how much contralateral hypertrophy can be expected from standard therapeutic-dose unilateral RE; however, reported figures vary between 20.3%[17] and 42%[14] and the mean observed contralateral volume increase in the RE group in our study (29%) falls well within this range, indicating that these results are not biased by any methodological particularity in performing RE at our institution and are likely to be reproduced in other centers. Furthermore, little is known about the kinetics of contralateral volume changes after RE but recent studies suggest that hypertrophy may take substantially longer to develop than after PVE and may even be a continuous process extending beyond 9 months after treatment.[14, 19] Therefore, it may be argued that RE may compare more favorably with PVE if more time is allowed between treatment and follow-up imaging. However, hypertrophy induction is relevant in patients who are candidates for surgery and there is a consensus that surgery should be performed as soon as resectability is achieved, because delaying surgery for several months to wait for maximum volume response carries the risk of tumor progression or requires extensive preoperative chemotherapy to eradicate disseminated tumor cells and micrometastases. Hence, for this very first study directly comparing contralateral hypertrophy induction with RE or PVE, an interval of ∼6 weeks between RE and follow-up imaging was considered appropriate. Although this was slightly longer than in the PVE cohort, leaving a patient without chemotherapy for 6 weeks is acceptable. Moreover, within the range of intervals used in our matched cohort, the time span between treatment and follow-up was not found to be an independent factor influencing contralateral volume response in the ANCOVA analysis (Table 2) and there was no significant correlation between the time from treatment to follow-up and the degree of contralateral hypertrophy observed in both groups, indicating that the slightly longer waiting time from treatment to follow-up in the RE group did not bias the comparison of contralateral hypertrophy induced by both treatments. Since we were able to demonstrate that RE induces substantial hypertrophy even at a 6-week interval, we feel that a prospective trial comparing PVE and RE a as means of hypertrophy induction is warranted. However, the endpoint of such a trial, which is currently being prepared at our institution, should be resection rate or even survival rather than mere volume increase, because hypertrophy induction is beneficial only if it leads to secondary resectability. Finally, contralateral volume increase after PVE may not always accurately mirror the gain in liver function. It has been demonstrated that FLR function as measured by mebrofenin hepatobiliary scintigraphy may increase to an even greater extent than FLR volume,[42] which may have an impact on clinical outcome.

In conclusion, our study demonstrated that unilateral RE with a standard therapeutic dose produces significantly less contralateral hypertrophy than PVE within a comparable time frame. However, the hypertrophy induced by RE is substantial and may be sufficient to achieve resectability in many patients, making RE a potentially valuable option for hypertrophy induction in specific situations.