Growth of colorectal liver metastases is not accelerated by intraportal administration of stem cells after portal vein embolization

Overview

Introduction: Future liver remnant volume (FLRV) is a crucial factor impacting resectability of colorectal liver metastases (CLM). In case of low FLRV, augmentation can be done by performing portal vein embolization (PVE). However, there is a risk of progression of CLM between PVE and resection. Intraportal application of autologous hematopoietic stem cells (HSC) is a possibility to accelerate the growth of FLRV. The effect of thus applied SC on CLM progression still remains unclear, though.

Methods: 63 patients underwent PVE between 2003 and 2015. In 20 patients a product with HSC was applied intraportally on the first day after PVE (PVE HSC group). HSC were gained from peripheral blood (10 patients) or bone marrow (10 patients). FLRV and volume of liver metastases (VLM) were evaluated by CT volumetry. The gained data were statistically evaluated in relation to the disease free interval (DFI), overall survival (OS), achievement of CLM resectability and progression of extrahepatic metastases. We compared the PVE HSC group with the group of patient undergoing simple PVE.

Results: No significant difference in FLRV and VLM growth was observed between the study groups. The percentage of exploratory laparotomies was smaller in the group with PVE and HSC application. Patients with simple PVE had a significantly higher incidence of extrahepatic metastases during follow up. We did not observe any significant differences in DFI and OS between the groups.

Conclusion: HSC application did not accelerate CLM growth in comparison with PVE alone. PVE and HSC application had a higher percentage of patients undergoing liver resection and a lower incidence of extrahepatic metastases.

Keywords:

colorectal metastases – portal vein embolization – Stem cells

Introduction

The development and progress of diagnostic me­thods, operative techniques, perioperative care and multidisciplinary approaches allow for planning as well as performing extensive liver resections in an increasing number of patients. In patients with colorectal liver metastases (CLM), resection is a fundamental cornerstone of treatment. Thanks to radical R0 liver resection, it is possible to achieve 5-year survival in more than 50% of patients [1].

CLM occur in 50% of patients being treated for colorectal carcinoma [2], and only 20−25% of these are primarily resectable [3,4,5]. In case the patients´ overall health status facilitates surgery, extensive liver resection (resection of three and more segments) is limited by the Future Liver Remnant Volume (FLRV) [6]. In patients undergoing chemotherapy, 30−35% of the FLRV must be left for good postoperative liver function. In cases of prolonged chemotherapy, patients with blue or yellow liver syndrome are recommended for up to 40% of FLRV [7,8,9].

Since it was first used by Makuuchi in the 1980s [10], portal vein embolization (PVE) has already become a standard method to increase the FLRV in primarily unresectable tumors [11,12,13,14]. FLRV growth occurs within 3−4 weeks after PVE, and the period between the PVE and resection is 2−60 days [13]. However, there is a risk of CLM progression while the FLRV is growing [15,16,17].

Stem cells (SCS) are a promising stimulator of FLRV growth in preclinical studies. The effects of mesenchymal stem cells (MSC) as well as hematopoietic stem cells (HSC) on liver regeneration were described in in vitro and in vivo studies [18]. A study using a large experimental animal showed a positive effect of MSC application into the portal vein so that the liver parenchyma regenerative potential could be increased [19]. Bone marrow SCS (BMSC) were used in the treatment of chronic liver failure in human medicine [20,21].

The use of HSC prevails in human medicine due to their abundance and relatively easy accessibility from the bone marrow or peripheral blood [19,22]. New studies dealing with the use of HSC in patients with inoperable, mostly primary, liver tumors were done [23,24,25,26,27]. Up to now, no study has been performed in a group of patients with CLM who would be monitored for malignant progression after performing PVE. The effect of HSC on CLM growth is still unclear, too.

The aim of this study is to evaluate the impact of HSC administration together with PVE on FLRV, the growth of FLRV and malignant progression between the procedure and liver surgery and to assess its impact on CLM operability in comparison with patients undergoing PVE alone.

Methods

This retrospective study includes patients with primarily inoperable CLM (treated at the institution of the first author between 2003 and 2015) who were designated as needing PVE. These include either patients with primarily inoperable CLM due to insufficient FLRV before PVE (less than 30%) or patients with an increased risk of postoperative failure as a consequence of prolonged oncologic therapy in combination with low FLRV (less than 40%). None of the patients had had any extrahepatic metastases before they were involved in this study. Patients were divided into two groups, the first being treated by PVE alone (Group PVE) and the second being treated by PVE with HSC application (Group PVE HSC). In addition to these basic groups, we selected patients with an initial FLRV less than 30% (Group PVE 30 and Group PVE HSC 30) from the basic groups for further analysis.

In the given period, this study included 63 patients. Before involvement in the study, the patients provided their informed consent. The study was approved by the Institutional Ethics Committee. In the study, there were 43 patients in Group PVE, and Group PVE HSC included 20 patients. Group PVE 30 consisted of 25 patients, and in Group PVE HSC 30, there were 8 patients. The study included 12 women and 51 men. Group PVE consisted of 9 women (26%) and 34 men (79%), and Group PVE HSC had 3 women (15%) and 17 men (85%). The average age was 62.2 years (33−79 years). The average age was 61.9 years in Group PVE (33−79 years) and 62.8 years in Group PVE HSC (50−75 years). The average age was 62.9 years in Group PVE 30 (46–79 years) and 64.4 years in Group PVE HSC 30 (51−75 years).

All patients underwent PVE via a percutaneous transhepatic method. A mixture of Histoacryl (BBraun, Tutlingen, Germany) and Lipiodol (Guerbert, Rennes, France) at a 1:10 ratio was used for embolization itself [7].

A product containing HSC was intraportally administered to the patients of Group PVE HSC within 24 hours after PVE was performed. HSC were collected in two ways. The first was HSC collection from the peripheral blood [24], and the other was the collection of SC from the bone marrow (BM) [26,27,28].

The first group consisted of 10 patients in whom HSC were collected from the peripheral blood. Before PVE, the patients underwent stimulation for HSC release from the bone marrow by subcutaneous application of granulocyte-colony stimulating factor (G-CSF, Neupogen, Filgrastim, Amgen Europe B.V., Breda, Holland) at a dose of 10 µg/kg/day. The stimulation took four days. On the fifth day, HSC apheresis and separation were performed using the Cobe Spectra continuous blood cell separator (BCT, Lakewood, CO, USA), and the HSC produced were preserved in the Anticoagulant Citrate Dextrose Solution (ACD-A, Baxter, Deerfield, IL, USA).

The other group consisted of 10 patients in whom HSC were collected from the bone marrow of the iliac crest under general anesthesia. Afterwards, the HSC were separated from the BM by centrifugation in the SynGenX™ Stem Cell Processing System (Thermo Fischer Scientific, Waltham, MA, USA) based on the weight gradients of the cells. The HSC were immediately applied into the non-occluded portal vein after centrifugation. BM collection and application of the obtained product were performed simultaneously under general anesthesia. The HSC product was obtained by means of apheresis and was also administered under general anesthesia. The administration procedure was the same in both groups. The patient´s abdominal cavity was accessed using a Gridiron´s incision. The venous arcade was prepared in the ileocecal area, and catheterization of the non-occluded portal bloodstream was performed by an interventional radiologist. A product containing HSC was applied into the non-occluded branch of the portal vein through the catheter.

The number of HSC (CD133+ and CD34+ cells) was evaluated by flow cytometry. Minimal amounts of HSC were administered at 1×10⁷ cells (HSC from peripheral blood) and 2.73×10⁸ cells (HSC from BM).

The effects of PVE and PVE with HSC application were evaluated by computed tomography (CT) with liver volumetry performed at the beginning of each week for three weeks after the procedure. Total liver volume (TLV), FLRV and Volume of Liver Metastases (VLM) were calculated. CT volumetries were performed using the Somatom Definition Flash (Siemens, Munich, Germany) and Somatom Sensation (Siemens, Munich, Germany) devices. CT volumetries were evaluated by the same team of radiologists using the semi-automatic software Syngo, via Liver Analysis (Siemens, Munich, Germany).

The CT volumetry data were compared between the PVE group and the PVE HSC group. Both groups were mutually compared with respect to overall survival (OS), disease-free interval (DFI), progression of extrahepatic metastases and the achievement of resection. Afterwards, the observed data were also evaluated in Group PVE 30 and Group PVE HSC 30.

The data were evaluated by statistical analysis. From the volumetric measurements of TLV, FLRV and VLM (in mL), the differences (3 weeks after PVE – before PVE; in mL) and rates of growth (difference/PVE duration; in mL/day) between the initial and the terminal values were calculated. From the FLRV and VLM, their proportions (in %) related to the appropriate TLV values were also calculated alongside the differences (% 3 weeks after PVE – % before PVE; in %) and rates of growth (difference of proportion/PVE duration, in %/week). All these indicators of tissue growth were subsequently analyzed. Two-sample comparisons were performed using the Mann-Whitney U-test. Correlations were analyzed using Spearman’s rank correlation with appropriate significance tests.

The DFI and OS in the patients who underwent resection were analyzed using the Kaplan-Meier method. Statistical significance of the differences in the DFI or OS between the two groups was assessed using the Gehan-Wilcoxon test. Point estimates of the observed survival times (or survival quantiles) were calculated from the Kaplan-Meier estimates of the survival functions using linear interpolation between the complete observations to the required time (or proportion surviving). The median follow-up estimates were calculated using the reverse Kaplan-Meier method [29].

All reported p-values are two-tailed, and the statistical significance level was set at α=0.05. Statistical processing and testing was performed using the STATISTICA data analysis software system (StatSoft, Inc. 2013. Version 12. www.statsoft.com.).

Results

Initial condition of the patients

We observed no statistically significant difference in the initial patient conditions before PVE among any of the groups, particularly concerning the age, FLRV and VLM before PVE (Graph 1A –1C, respectively). The average FLRV before PVE was 29.2% (15.1−47.3%, median 28.6%) in Group PVE and 30.2% (15.3–45.6%, median 30.7%) in Group PVE HSC. The average VLM before PVE was 170.6 mL (5–2232 mL, median 50 mL) in Group PVE and 66.4 mL (3.4–465 mL, median 31.5 mL) in Group PVE HSC.

Growth of FLRV

Three weeks after PVE, FLRV was not statistically significantly different between Group PVE and Group PVE HSC. Average FLRV was 36.4% (18.9–53.3%, median 35.8%) in Group PVE and 39.6% (28.3–51.8%, median 41.8%) in Group PVE HSC (Graph 2A) after 3 weeks. The average growth of FLRV proportion was not significantly different between these monitored groups within 3 weeks after PVE, either. The average growth was 7.2% (-7.0–24.7%, median 6.0%) in Group PVE and 9.4% (-5.8–22.4%, median 10.7%) in Group PVE HSC (Graph 2B). In absolute terms, average FLRV growth was 121.2 mL (-173.0–480.0 mL, median 122 mL) in Group PVE and 157.4 mL (-35–340 mL, median 181 mL) in Group PVE HSC (Graph 2C). This difference was also statistically non-significant. The same holds for the average rate of FLRV increase, which was 2.9 mL/day (-6.4–14.3 mL, median 2.4 mL) in Group PVE and 3.4 mL/day (1.3−13.1 mL/day, median 2.7 mL/day) in Group PVE HSC (Graph 2D).

Progression of CLM

In 5 of the patients (11.6%) of Group PVE and no patients from Group PVE HSC, surgery was cancelled due to a clear progression of malignant disease found on the CT (bilobar liver metastases growth and new lung and peritoneal metastases). In the case of patients after PVE with HSC, exploratory laparotomy was performed in 3 of 20 patients (15%), and in the case of PVE only, this was done in 17 of 43 patients (39.5%). However, this result is not statistically significant. A statistically significant factor impacting the contraindication of resection was the progression of metastases (p=0.044) (Graph 3A). However, no statistically significant difference in the progression of metastases could be seen between Group PVE and Group PVE HSC. In 3 weeks after PVE, the average VLM was 261.1 mL (7.0–2500 mL, median 91.5 mL) in Group PVE and 107.9 mL (5.0–832.0 mL, median 64.5 mL) in Group PVE HSC (Graph 3B). In 3 weeks after PVE, the average growth of the VLM was 81.1 mL (-29.0–800.0 mL, median 30.6) in Group PVE and 41.5 mL (-28.0–367.0 mL, median 12.5 mL) in Group PVE HSC (Graph 3C). The average rate of VLM increase was 2.0 mL/day (-0.5–20.5 mL/day, median 0.8 mL/day) for PVE and 0.9 mL/day (-0.8–8.2 mL/day, median 0.4 mL/day) for PVE HSC. Between the compared groups, there was a statistically significant difference related to the occurrence of extrahepatic metastases, which was greater in Group PVE compared to Group PVE HSC (p=0.0210) (Graph 3D).

DFI and OS analysis

In case of resected patients, DFI was not statistically significantly different between Group PVE and Group PVE HSC. In Group PVE, DFI was 25.6% (1^st year), 2.0% (3rd year) and 0% (5^th year). The median DFI was 0.445 years with PVE. In Group PVE HSC, the DFI was 23.2% (1^st year), the estimated DFI was 5.8−10.5% (3^rd year) and the estimated DFI was 0.9-10.5% (5^th year). The median DFI was 0.367 years with PVE HSC (Graph 4A).

The OS in patients after resection was not statistically significantly different between Group PVE and Group PVE HSC. In Group PVE, the median was 3.303 years. In Group PVE, the OS was 84.5% (1^st year) and 59.4% (3^rd year) and the estimated OS was 6.7−14.7% (5^th year). In Group PVE HSC, the median was 1.412 years. The OS for Group PVE HSC was 81.3% (1^st year), estimated OS 29.9−49.9% (3^rd year) and estimated OS 4.7−49.9% (5^th year) (Graph 4B). According to clinical data, we can state that only 6 of the 40 patients of Group PVE survived for more than 3 years (1 patient for 3 years, 3 patients for 4 years, 1 patient for 5 years and 1 patient for 6 years). In Group PVE HSC, 2 patients survived for more than 3 years (1 patient for 4 years and 1 patient for 6 years). This last patient was the only one still alive that did not have any recurrence.

Significantly faster recurrence can be seen in patients without adjuvant chemotherapy in the whole sample (Groups PVE and PVE HSC combined, p=0.0086) (Graph 4E); however, this effect is not statistically significant in Group PVE or Group PVE HSC.

Results for Groups PVE 30 and PVE HSC 30

The initial conditions of the patients (age, initial FLRV, initial VLM) were not statistically significantly different between Group PVE 30 and Group PVE HSC 30 (Graph 1D–1F, respectively). The average FLRV before PVE was 25.2% (15.3%−29.5%, median 26.1%) in Group PVE 30, and in Group PVE HSC 30, it was 22.8% (15.3–29.5%, median 21.7%). The average VLM before PVE was 140.0 mL (5.0–1043.0 mL, median 57 mL) in Group PVE 30 and 37.6 mL (3.4–97.0 mL, median 31.5 mL) in Group PVE HSC 30.

Growth of FLRV

Three weeks after PVE, the average FLRV was not statistically significantly different between Group PVE 30 and Group PVE HSC 30. The FLRV was 34.5% (18.9–53.3%, median 34.7%) in Group PVE 30 and 36.3% (28.3–43.1%, median 37.0%) in Group PVE HSC 30 3 weeks after PVE (Graph 2E). The average increase in the FLRV proportion was 9.3% (-2.2–24.7%, median 8.4%) in Group PVE 30 and 13.5% (2.3–22.4%, median 12.7%) in Group PVE HSC 30 (Graph 2F). The average rate of FLRV increase was 4.2 mL/day (-6.4–11.7 mL/day) in Group PVE 30 and 5.0 mL/day (1.7–13.1 mL/day) in Group PVE HSC 30.

Progression of CLM

Three weeks after PVE, the average VLM was 252.7 mL (18.0–1843.0, median 140.5 mL) in Group PVE 30, and in Group PVE HSC 30, it was 60.5 (5.0–110.0, median 64.2 mL) (Graph 3E). The average growth of the metastases within 3 weeks after PVE was performed was 99.6 mL (-13.0–800.0 mL, median 46.4 mL) in Group PVE 30 and 23 mL (0.0 –78.1 mL, median 11.5 mL) in Group PVE HSC 30 (Graph 3F). The average growth rate of VLM was 2.4 mL/day (-0.4−20.5 mL/day, median 0.9 mL/day) in Group PVE 30 and 0.4 mL/day (0–1.2 mL/day, median 0.3 mL/day) in Group PVE HSC 30. The observed differences were not statistically significant.

DFI and OS analysis

There was no statistically significant difference related to the DFI between the monitored groups. DFI was 19.7% (1^st year) in Group PVE 30. It was impossible to define the DFI for the third and fifth year. In Group PVE HSC 30, the DFI was 23.8% (1^st year), the estimated DFI was 12.3−22.5% (3^rd year) and the estimated DFI was 1.9–22.5% (5^th year) (Graph 4C). The OS was not significantly different between the monitored groups. The OS related to Group PVE 30 was 74.5% (1^st year) and 35.5% (3^rd year), and the fifth year could not be defined. The OS related to Group PVE HSC 30 was 63.5% (1^st year), the estimated OS was 21.4−35.7% (3^rd year) and the estimated OS was 3.3−35.7% (5^th year) (Graph 4D).

Discussion

FLRV is a principal factor that influences the resectability of CLM. Low FLRV entails a risk of acute liver failure in patients after large liver surgery. FLRV may be increased by performing PVE, portal vein ligation or so-called ALPPS (Associating Liver Partition and Portal Vein Ligation for Staged Hepatectomy). ALPPS results in a faster increase of FLRV, in comparison with PVE [30]. However, this method still has higher morbidity and mortality [9,31,32,33]. Application of HSC in the non-occluded branch of the portal vein can possibly accelerate the FLRV growth after PVE. Nevertheless, there is a risk of the progression of malignancy [34]. In our study, we monitored the effect of the application of HSC after PVE on FLRV growth, VLM growth, the pre­valence of extrahepatic metastases, and liver resection achievement as well as on the DFI and OS after liver surgery for CLM.

The groups of patients compared in this study were identical to each other regarding age, FLRV and VLM before PVE. Before PVE, the patients had no extrahepatic metastases. The study groups showed no statistically significant differences in terms of FLRV growth. How­ever, the figures and final data may show a trend toward a higher rate of FLRV growth in patients with applied HSC. This trend was most pronounced in Group PVE HSC 30. The FLRV figure before PVE shows that Group PVE 30 had a lower FLRV in comparison with Group PVE HSC 30. However, the FLRV became higher 3 weeks after PVE with the administration of HSC. This trend was also illustrated by the average growth of FLRV that was increased by 4% in Group PVE HSC 30, compared to Group PVE 30. Faster growth of FLRV was also described in previous studies using HSC for the stimulation of FLRV growth [23,24,25,26,27].

In general, there are three ways in which SC can influence liver regeneration. These include transdifferentiation, where the SC in damaged liver are differentiated into target cells, and fusion, where multinuclear regenerating hepatocytes appear by the SC merging with dama­ged liver cells [35]. The last theory has to do with the effect of paracrine stimulation in liver cells, where the SC create a suitable microenvironment for the regeneration of hepatocytes [36,37]. The question is whether SC and their application into the portal vein can cause stimulation of the tumor [37]. None of the studies mentioned above dealt with malignant progression after PVE and HSC application. CT scans and volumes of metastases show an obvious metastatic progression in all the study groups. Nevertheless, since the difference in the VLM growth was not statistically significant between the compared groups, HSC application apparently did not lead to the stimulation of tumor progression.

The combination of slow metastatic progression and accelerated FLRV is absolutely crucial for patients because they have a higher chance of undergoing radical liver resection and the percentage of exploratory surgeries is thus reduced. In our patients, the observed percentage of exploratory laparotomies was considerably lower in Group PVE HSC than in Group PVE, suggesting a distinct therapeutic benefit of HSC application for patients.

It is worth noting that the progression of extrahepatic metastases is significantly higher in groups with simple PVE. The effect of PVE on progression of the tumors was mentioned above. However, the principle of the interaction between the BMSC and tumor tissue is still unclear. BMSC secrete several cytokines, chemokines and growth factors, which are potentially disease-modifying by influencing angiogenesis, suppressing inflammation or inhibiting apoptosis [36]. Potential development or stimulation of tumor growth by SC was described in pancreatic cancer [38]. There are theories of tumorigenesis suggesting that cancer stem cells support the metastatic disease and spreading of tumors. These theories disrupt the classical step-by-step theories of tumorigenesis. Primary tumors could have the ability to establish a suitable microenvironment for their dissemination even in distant tissues. It is supposed that HSC might support the preparation of this pre-metastatic microenvironment [39]. In this context, HSC application into the portal vein could bring a risk of tumor dissemination into the liver as well as other organs, e.g., into the lungs. On the other hand, there was a case report study stating that the administration of CD133+ autologous stem cells did not negatively affect the biology of the disease [8].

Additionally, our study did not prove a negative effect of intraportal administration of HSC after PVE on the progression of malignancy. In contrast, extrahepa­tic dissemination was statistically significantly worse in the group with PVE alone. At the moment, we are not able to explain this effect in more detail. The limited extent of our patient sample may be one of the reasons; therefore, we shall continue in our project to obtain more data.

The evaluation of DFI and OS is difficult in our small group, especially considering the small number of patients that survived for more than 3 years. Therefore, we must present estimated DFI and OS in some cases. Between the compared groups, there were no statistically significant differences in DFI or OS. Most patients had an early recurrence. According to the facts mentioned here, the patients who were included in this study had primarily non-resectable CLM and had a very poor prognosis. Our study also confirmed the positive effect of oncologic therapy after liver resection in the patients after FLRV growth stimulation. This illustrates that using radical resection combined with oncologic therapy is the best possible method for treating CLM.

Conclusion

Our study has shown that HSC application did not accelerate CLM growth after PVE in comparison with PVE alone. The positive effect of HSC application on the growth of FLRV observed in our sample was not statistically significant. However, the group where HSC application was performed after PVE showed a higher percentage of patients who underwent resection and a lower incidence of extrahepatic metastases. The combination of PVE with administration of autologous SC is particularly suitable for patients with primarily non-resectable CLM and FLRV lower than 30%.

Study was supported by: The National Sustainability Program I (NPU I) Nr. LO1503 provided by the Ministry of Education Youth and Sports of the Czech Republic

Centre of excellence UK/MED/006 „Centre for experimental and clinical liver surgery

Conflict of interests

The authors declare that they have not conflict of interest in connection with this paper and that the article has not been published in any other journal.

MUDr. Jan Brůha

Department of Surgery and Biomedical Centre

Faculty of Medicine and Teaching Hospital in Pilsen

alej Svobody 80

304 60 Pilsen

e-mail: bruhaj@fnplzen.cz