Dosimetric parameters predicting contralateral liver hypertrophy after unilobar radioembolization of hepatocellular carcinoma

Radioembolization (RE) using ⁹⁰Y–loaded microspheres is increasingly used in a palliative setting for primary [1, 2] and secondary liver diseases [3]. The interest of RE for downstaging has also been described [2, 4, 5] especially in a recent randomized phase 2 study highlighting the superiority of RE in comparison with chemoembolization [5] opening new perspectives of surgical ablation for patients with large tumors.

Several retrospective studies have recently shown the capacity of RE to induce, after unilateral treatment, a significant hypertrophy of the opposite lobe [6‐13], which is of particular interest in a neoadjuvant setting. Indeed, it is recognized that for patients with underlying liver disease, the hepatic reserve or proportion of the future remnant liver (FRL) following RE should be higher than 40 to 50% of the initial liver mass to avoid post hepatectomy liver failure [12, 14].

Historically, portal vein embolization (PVE) was used to increase the FRL of patients without cirrhosis (mainly with metastatic disease). RE has the advantage of having a direct therapeutic effect on tumors and has been recently proven to be effective in cirrhotic HCC patients [7, 12, 13]. Therefore, understanding the factors inducing liver hypertrophy after RE and developing RE for hypertrophy purposes is of major interest.

The reported mean of hypertrophy after RE varies from 29 to 45% depending on the studies and on the evaluation time. It is now recognized that the RE-mediated hypertrophy takes place over several months [13, 15, 16] and that hypertrophy is still increasing between 9 and 12 months after RE [15]. In patients treated by RE, the factors related to hypertrophy are not well described, especially dosimetric factors have not been evaluated to date except the injected activity [10] or the mean treated liver dose [12, 13, 16] without any statistically significant link identified.

Due to a minimally embolic effect in comparison with PVE, radio-induced damage and cell death within the treated liver lobe are likely to be the main inducer of the hypertrophy in the contralateral RL after RE. Thus, we hypothesized that RL hypertrophy could be at least in part in relation to the levels of injuries and induction of cell death in the injected healthy liver and/or of the tumor itself possibly in relation to the production of growth factors and cytokines [17‐20]. It is mandatory to keep in mind that the threshold doses inducing damages are different in tumor versus healthy liver.

In this context, two dosimetric parameters are particularly interesting to evaluate the injected healthy liver dose (HILD) and the tumor dose especially for large lesions as recent studies have demonstrated the accuracy of MAA-based dosimetry in the prediction of tumor response [21‐23] and toxicity [24, 25] for HCC patients.

The goal of this study is to evaluate the ^99mTc-labeled MAA dosimetric parameters prior HCC setective internal radiotherapy that would be associated with elevated contralateral hypertrophy after RE with the objective in the future to define specific dosimetric endpoints predicting liver hypertrophy.

In this report, we have measured the dosimetric parameters following unilobar injection of ^99mTc-labeled MAA in patients suffering from HCC and prior ⁹⁰Y–loaded microspheres radiotherapy in order to define specific dosimetry endpoints associated with elevated contralateral hypertrophy in non-injected liver after RE.

The first main result of this study is the clear impact of the HILD on the occurrence of a MHT ≥ 10% identified both at univariate (p = 0.0081) and multivariate analysis (p = 0.0029). MHT ≥ 10% was observed in 92.2% of the patients having a HILD ≥ 88Gy, (representing 52% of the population), while the MHT ≥ 10% was found in only 65.7% of the patients when HILD was <88 Gy. To the best of our knowledge, this is the first demonstration that a dosimetric parameter positively correlates with hypertrophy. Indeed, previous studies have failed to find any correlation between dosimetric parameters and hypertrophy. However, in these reports, the dosimetric evaluation relied on the dose to the treated liver [12, 13], which is not the best dosimetric parameter to evaluate since it is a mean dose between TD and HILD. This has also been the case in the study previously reported by Fernandez-Ros et al. [18] who have evaluated the HILD using the partition model; however, it seems that in this study HILD was only evaluated as a continuous variable. The fact that radio-induced damage is determinist and responds to a threshold can explain the absence of a link between HILD and hypertrophy in the study published by Fernandez-Ros and colleagues [18].

One way to significantly enhance hypertrophy seems to be to deliver HILD higher than 88 Gy. However, the maximal deliverable HILD without inducing liver failure is not well described. Chiesa et al. [24] found a mean healthy liver dose (mean dose to the injected and not injected healthy liver) of 75 Gy producing a probability of G2 liver toxicity of 15%. In a previous study of 71 cases, we found no link between HILD and liver toxicity but only a significant link for HILD >100 Gy and a low hepatic reserve (<30%) [25]. In the present study, no significant link was found between liver G3 permanent toxicity and HILD (<88 versus ≥88Gy). However, the maximal HILD to deliver to induce the optimal hypertrophy in case of low initial RL should be evaluated prospectively.

For the patients with a low HILD (<88 Gy, i.e., 48% of the cases), other dosimetric parameters than HILD, such as TD for large lesions, could have an impact on hypertrophy. In this situation MHT ≥ 10% was present in 70.8% of the patients with a high dose to large tumors (TD_{≥ 205Gy for TV ≥ 100cm} ³) as against only in 54.4% for low tumor dose (<205 Gy) or small tumor volume (<100cm³). This result was not statistically significant, but this trend highlights the potential impact of high tumor dose for large lesions on MHT. However, in one subgroup analysis (the subgroup of patients with a HR  < 50%) the parameter TD _{≥ 205Gy for TV ≥ 100cm} ³ significantly impacted the contralateral hypertrophy. In fact, in this subgroup of patients, the MHT mean was significantly higher in cases of TD _{≥ 205Gy for TV ≥ 100cm} ³ (62.3 ± 53.3%) versus only 29.1 ± 25.6% in cases with a TD _{< 205Gy} or a TV < 100 cm³, (p = 0.0329). This group of patients is of particular importance, as it represents the population that most needed the hypertrophy to be induced by RE.

It is interesting to underline that, for the same injected liver dose the lower the HILD is, the larger usually the tumors are with high doses due to the partitioning of the activity between the healthy liver and tumor. So for patients with low HILD, it seems possible to safely increase the dose in cases of large tumors, to obtain TD ≥ 205Gy, in order to stimulate hypertrophy, without reaching too high a HILD.

Finally, MHT ≥ 10% was seen in 83.9% for patients with HILD ≥ 88 Gy or a TD_{≥ 205Gy for TV ≥ 100cm} ³, representing 85% of the population, while MHT ≥ 10% was found only in 54.5% (p = 0.0265) for patients with none of those parameters (p = 0.0265).

Thus, our results are suggesting that optimizing and increasing either the HILD or the TD for large lesions could result in producing significantly higher hypertrophy for at least 85% of the patients. This data should be taken into account when discussing RE in the neoadjuvant setting of HCC. One issue not evaluated in this study is the potential impact of increasing the number of microspheres used instead of the doses. Other studies are warranted to evaluate the potential impact of microspheres number (i.e., the embolic load) on hypertrophy after RE.

After loss of the hepatic mass, the liver/body weight ratio returns to 100% because of the ability of the liver to regenerate. This process, also called Hepatostat [26, 27], allows the adjustment of its size to the overall body metabolic requirement. Using both in vitro and in vivo rodent models, it has been shown that the unique ability of differentiated hepatocytes to exit from quiescence and to proliferate is controlled by a plethora of cytokines and growth factors. The pro-inflammatory cytokines tumor necrosis factor α (TNFα) and interleukin 6 (IL-6) are the early stimuli during the liver regeneration allowing the priming of hepatocytes, whereas growth factors such as HGF, TGFα and EGF regulate the onset of DNA synthesis [28‐30]. Many other soluble factors such as the vascular endothelial growth factor (VEGF), platelet derived growth factor (PDGF), and Augmenter of Liver Regeneration (ALR) strongly potentiate the effects of the pro-inflammatory cytokines and growth factors to achieve a complete liver regeneration and to rescue any partial defect in the secretion of one or another factor.

While the molecular mechanisms triggering the liver regeneration after partial hepatectomy have been extensively studied in rodents, the growth factors and downstream pathways regulating proliferation of hepatic cells in humans during chronic liver diseases are not well understood. In diseased human liver, the chronic loss of mature hepatocytes is not compensated by an efficient proliferation leading to the development of cirrhosis, the alterations of the liver macrophage phenotype, the abnormal production of cytokines and the appearance of progenitor cells [27]. In this pathological context, the surgical resection or the RE of HCC induce a more or less efficient liver regeneration of the non-tumoral liver although the humoral mediators remain poorly described.

However, a production of several cytokines and growth factors including VEGF, bFGF, PDGF, IL6 and HF has been recently demonstrated after RE that may constitute possible prognostic factors for the overall survival outcome [17‐20]. To date, the mechanisms that trigger this production and the cell types secreting these mediators during RE are not known.

Individualizing two targets, the healthy treated liver and the tumor itself, as done in this study, seems logical in case of RE, which is responsible of tumor necrosis and can lead to injuries in the healthy liver [31]. Then, several hypothesis concerning the biological mechanisms that trigger the liver hypertrophy can be proposed. Since RE is recognized to be minimally embolic, a portal flow redistribution from the treated liver to the RL or shear-stress that could induce hypertrophy after PVE [32] is unlikely to be the main initial event with RE, even though RE is known to produce variable signs of portal hypertension with time [8, 9].

A reduction of the functional mass of the treated liver, supposedly correlated with the HILD, can be a valuable hypothesis since a significant atrophy of the treated liver is described in all RE studies [6‐13]. This is the rationale for the concept of radiation lobectomy used by several authors [7, 13, 19]. In this situation, the treated lobe atrophy has been proven to be larger than the tumoral volume, confirming an atrophy of the healthy injected liver. The significant link between a HILD ≥ 88Gy and the frequency of HTM ≥ 10% found in the present study supports this hypothesis. The good tolerance of radiation lobectomy reported in all studies [6‐13] in case of insufficient FLR, in comparison with surgical lobectomy, which is an immediate ablative approach, is certainly in relation to the fact that maximal atrophy is time dependent and takes several months to occur [13, 15, 16]. Therefore, the atrophy taking place after RE is certainly compensated by the occurrence of liver regeneration in the FRL in the meantime.

A second possible mechanism is the direct antitumoral effect of RE, not seen with PVE, with the potential release of mediators in case of tumor stress and necrosis (supposed to be in relation with the TD). The amount of the released mediators certainly depends upon the tumor volume and the TD. Elsewhere, HCCs are recognized to be potentially functional and, in this situation, the destruction of HCCs can lead to a reduction of the functional mass of the liver. This hypothesis is supported by the trends and significant link in one subgroup between a TD≥205Gy for TV ≥ 100cm³ and the frequency of HTM ≥ 10% shown in our study.

Regarding global results of hypertrophy, the MHT mean reported herein after RE, i.e., 35.4 ± 40.4%, are in agreement with previously reported data, i.e., mean (or median) MTH ranking from 29 to 45% depending on the studies and on the evaluation time [13, 15, 16].

It is now clearly established that RE significantly reduces the amount of patients with inaccurate FLR. Indeed, it has been demonstrated in previous studies that the proportion of patients with a too small FLR significantly decreased after RE. A FLR < 50% decreased from 52.9% to 32.4%, p = 0.013, after RE in a previous study, p = 0.013 [12] and a FLR < 40% also significantly decreased from 56.6% to 29.4% after RE in the study published by Fernadez-Ros et al., p < 0.001 [16]. Similarly in our study, the proportion of patients with a FLR < 50% prior RE, significantly decreased from 47.9% to 13.6% only at time of MHT (p < 0.0001).

Thus one subsequent key question is “how long we could wait for hypertrophy occurrence after RE” as it is also recognized that in this situation hypertrophy takes more time to occur than with PVE. In our study DCR was 98.6% at 3 months underlying the absence of significant risk of progression within 3 months after RE. However, DCR reduced to 71.2% at 6 months (but with progression in the treated liver in only 10.9% of the patients and in the non-treated liver for 17.8% of the patients).

So, we can recommend to wait at least 3 months after RE to evaluate hypertrophy and to check for patients operability. At 3 months, if hypertrophy is not sufficient enough, later evaluations can be recommended, especially at 6 months, as hypertrophy can take more than 3 months to occur, but in this situation a significant amount of patients will no more be candidates for surgery due to tumor progression in the non treated liver (17.8% in our experience).

Those data are confirming the interest in RE to produce significant liver hypertrophy, even in patients with underlying cirrhosis, while producing in the same way an accurate tumor control, which is not achievable while using PVE.

The main limitations of this study are its retrospective nature and the number of patients included. Despite a relative large number of patients (i.e., 73) in comparison with previously published studies [13, 16] of post RE hypertrophy in HCC (52 and 67 HCC patients in the two largest previous cohorts) subgroup analysis power is certainly not sufficient since the number of subgroup cells was below ten for the several analysis.

Elsewhere in the evaluation of the FRL after RE, the volume of the treated liver is considered as functional as the FRL itself, which is certainly not the case. Then, a simple volumetric analysis, in this situation of RE, more than likely underestimates the real percentage of the function of the FLR. In fact, an important reduction of the liver function of the treated lobe has been recently described using hepatobiliary scintigraphy in two patients out of three treated by RE [33]. In the future, the use of hepatobiliary scintigraphy should be used to evaluate more accurately the hypertrophy post RE with a more functional approach than a volumetric approach alone.

This study demonstrates for the first time that HILD (< or ≥88 Gy) is significantly associated with liver hypertrophy (MHT ≥10%) with univariate and multivariate analysis underlying the robustness of this parameter. There is also a trend of the impact of high tumor dose in large lesions (TD _{≥ 205Gy for TV ≥ 100cm} ³) on MHT with a significant impact in one subgroup analysis. Larger prospective studies evaluating MAA dosimetric parameters have to be designed to confirm those results with the aim to finally define safe specific dosimetric endpoints favoring hypertrophy occurrence.