Portal Application of Autologous CD133+ Bone Marrow Cells to the Liver: A Novel Concept to Support Hepatic Regeneration



The liver has a large capacity for regeneration after resection. However, below a critical level of future liver remnant volume (FLRV), partial hepatectomy is accompanied by a significant increase of postoperative liver failure. There is accumulating evidence for the contribution of bone marrow stem cells (BMSCs) to participate in liver regeneration. Here we report on three patients subjected to intraportal administration of autologous CD133+ BMSCs subsequent to portal venous embolization of right liver segments, used to expand left lateral hepatic segments as FLRV. Computerized tomography scan volumetry revealed 2.5-fold increased mean proliferation rates of left lateral segments compared with a group of three consecutive patients treated without application of BMSCs. This early experience with portovenous application of CD133+ BMSCs could suggest that this novel therapeutic approach bears the potential of enhancing and accelerating hepatic regeneration in a clinical setting.


In up to 45% of patients with primary or secondary liver tumors, extended hepatectomy (greater than five segments) is necessary to achieve tumor-negative resection margins [15]. However, patients with an anticipated future liver remnant volume (FLRV) below 20%–25% of total liver volume (TLV) have an increased incidence of postoperative morbidity and mortality [58]. The concept of preoperative expansion of the left-lateral FLRV (segments II and III) using selective portal venous embolization (PVE) of contralateral liver segments I and IV to VIII before right trisegmentectomy is increasingly performed as a safe and effective concept to provide a proliferation stimulus (Fig. 1; cartoon on liver anatomy) [8, 9]. However, patients eligible for extended liver resection, such as those presented here, frequently suffer from large and fast progressing liver lesions, limiting the waiting time after PVE (observed to be up to 150 days [9]) to reach an adequate left lateral hepatic mass. Consequently, in some of these patients, PVE alone is insufficient to induce adequate proliferation of the left lateral FLRV in time for safe oncologic, extended liver resection.

Figure Figure 1..

Image outlining the distribution of hepatic segments. Parenchymal dissection along the falciform ligament as performed in extended right hepatectomy (trisegmentectomy) is shown.

To further accelerate liver proliferation levels, we consequently followed accumulating evidence for the contribution of extrahepatic stem cells (SCs) like hematopoietic progenitor cells participating in the concert of liver regeneration [1013]. Bone marrow (BM) cells have been shown experimentally to participate in liver proliferation after hepatic resection [14]. Furthermore, it has been postulated that hematopoietic progenitor cells are able to transdifferentiate into both hepatocytes and bile duct cells [11, 1519]. Mobilization of peripheral hematopoietic, CD34+ SCs (known to bear the capacity for differentiation into a hepatic lineage) after liver resection in oncologic patients has been demonstrated and was 10-fold higher compared with liver-sparing abdominal surgery [20]. These data indicate a possible role for BM-derived SCs in liver proliferation after substantial loss of liver mass. Although as yet inconclusive, the therapeutic potential of hematopoietic SCs seems to be an attractive prospect for liver repair after acute or chronic hepatic injury [21, 22].

CD133+ SCs were therapeutically used previously to support tissue and organ regeneration. Our center [23, 24] and others [25, 26] have used intramyocardial application of hematopoietic SCs enriched for CD133+ cells in various settings to promote the regeneration of postinfarction myocardium. In this study we report for the first time the therapeutic application of BM-derived SCs in humans with the intent to promote liver regeneration processes. CD133+ SCs highly enriched from autologous BM were selectively implanted to the left-lateral portal branches in three patients subsequent and in addition to selective PVE of contralateral liver segments I and IV to VIII.

Patients and Methods

Case Reports

Characteristics of the three patients we report on are summarized in Table 1. These patients, all characterized by large central liver malignancies, were selected for PVE before scheduled extended right hepatectomy, because FLRV ranged below 20% of TLV. However, PVE as an isolated measure was questionably sufficient to induce adequate proliferation of the left lateral FLRV within a time frame (observed to be up to 150 days after PVE in our own experience [9]) required for safe extended liver resection with tumor-free resection margins. Reasons were limited quality of hepatic parenchyma, known to impair regeneration potential [27] (patient 1), an unusual low basal volume of segments II and III as FLRV (patient 2), or a fast progressing liver lesion (patient 3).

Table Table 1.. Patient and treatment characteristics
  1. a

    Change in liver volume subsequent to PVE with application of CD133+ BM cells to portal branches of left lateral segments II and III (SCT) and without stem cell application (CTR), respectively, are stated. Characteristics for readministered cell fraction prepared from autologous BM are demonstrated for SCT patients.

  2. b

    Abbreviations: BM, bone marrow; CTR, control/PVE only; IDDM, insulin-dependent diabetes mellitus; PVE, portal venous embolization; SCT, PVE + stem-cell treated; TLV, total liver volume.

 SCT patient 1SCT patient 2SCT patient 3CTR patient 1CTR patient 2CTR patient 3
Patient characteristics 
    Age (yr)/gender47/m71/f53/f64/m65/f75/m
    Diagnosis of hepatic malignancyHepatocellular carcinomaNeuroendocrine metastasisColorectal metastasisHepatocellular carcinomaCholangiocarcinomaColorectal metastasis
    Maximum hepatic tumor expansion (cm)
    ComorbidityChronic replicating hepatitis C, hepatic fibrosis, no cirrhosisPrimary neuroendocrine tumor of the vermiform appendix (simultaneously resected)History of sigmoidal carcinoma with sigmoidal resection and four courses of 5-fluorouracil and folic acid 10 mons before IDDM, compensated kidney insufficiency, severe anorexiaArterial hypertension, hypercholesterinemia, no hepatic fibrosis nor cirrhosisNoneChronic gastritis
    TLV before PVE (ml)2290.01601.01360.01702.01940.01276.0
    Volume: left lateral lobe before PVE (ml)456.0193.0260.0337.0438.0330.0
    Absolute gain of left lateral lobe volume (ml)232.0235.0111.093.0102.090.0
    Relative gain of left lateral lobe volume (%)50.9121.842.728.023.327.0
    Left lateral lobe volume (% of TLV before PVE)19.912.119.119.822.625.9
    Left lateral lobe volume (% of TLV after PVE)30.026.727.325.327.831.4
    Time for gain of left lateral liver volume (day)
    Daily gain of left lateral liver volume (%/TLV)0.4590.6950.5860.250.2260.212
    Daily gain of left lateral liver volume (ml/day)10.611.
Characteristics of processed BM cells (cytofluorimetric analyses) 
    Aspirated total BM volume (ml)220.0220.060.0
    CD133+ purity of cells for application (%)91.011.969.7
    Absolute number of CD133+ cells applied8.812.32.4

Reference Group

A group of three consecutive non-BMSC patients, treated with the same therapeutic protocol, except for the application of SCs, served as a nonrandomized reference. Latter patients also suffered from large central hepatobiliary malignancies, fulfilling criteria for presurgical expansion of the FLRV like the patients presented here selected for BMSC application. Reference patients received an intervention by the same radiologists during the same time span as those three receiving BMSC treatment subsequent to PVE. Non-BMSC patients were free from both cholestatic disease and a history of chronic liver disease.

Liver Volumetry

Data sets for volumetry were obtained from helical computerized tomography (CT) of the upper abdomen both before segmental occlusion of the portal vein and before extended hepatectomy TLV (calculated hepatic volume minus tumor volume) and FLRV (segments II and III) were calculated based on these data sets, obtained in the portal vein enhancement phase. For the documentation of interobserver agreement, volumetry was calculated by a second observer, who was blinded to patient identity and the results of the first observer. Average interobserver variability was 1.25% (0.45%–2.2%).

Portal Venous Embolization

For PVE, a transileocolic portal venous approach via direct cannulation under general anesthesia was used in all patients. Subsequent to placement of a 5-F vascular sheath (Terumo, Leuven, Belgium) in the portal vein that was exteriorized through the abdominal wall incision, fascia and skin were closed. Flush portographies were performed with a 5-F angiographic cobra catheter (Terumo, Leuven, Belgium) in the main portal vein. Polyvinyl alcohol particles (Contour, Boston Scientific, Cork, Ireland) ranging from 300–1,000 μm, microcoils (VortX, Boston Scientific), and a Histoacryl/Lipiodol mixture (Braun, Tuttlingen, Germany and Guerbet, Roissy, France) were used for selective occlusion of portovenous branches to liver segments I and IV to VIII. Successful embolization was documented by a final flush portography.

Preparation and Characterization of CD133+ Bone Marrow Cells

In accordance with institutional ethical regulations and after obtaining patients' informed consent, the whole procedure from harvesting BM until readministration of selected cells was performed in a closed system. Autologous BM, 220 ml (patients 1 and 2) and 60 ml (patient 3), aspirated from the posterior iliac crest were drawn in heparin-coated syringes after the induction of anesthesia right before minilaparotomy for portal catheter placement was performed. BM cells were prepared simultaneously with the PVE procedure. The cell suspension was filtered to remove bone spicula and then processed by a GMP-grade cell-selection unit (CliniMACS, Miltenyi Biotec, Bergisch Gladbach, Germany) to enrich for CD133+ cells as previously described [23, 28]. After an average of 160 minutes, the enriched mononuclear cells were ready for intrahepatic application. Cells were resuspended in a total volume of 80 ml phosphate-buffered saline solution.

Aliquots from the BM aspirate and the injected cell fraction were collected for FACS analysis as reported previously [23, 24, 28]. The number of mononuclear cells was determined by a cell counter (Sysmex, Duesseldorf, Germany).

Application of CD133+ Bone Marrow Cells

After completion of PVE, CD133+ cells were selectively applied to the nonoccluded portal branches of liver segments II and III as equal aliquots of 40 ml, taking an average time of 4 minutes each. This application was performed using a 5-F cobra catheter (Terumo, Leuven, Belgium) introduced into the segmental branches II and III under fluoroscopic guidance. After intrahepatic placement of cells, the portal catheter and the sheath were removed.

Statistical Analysis

The increase in liver volume in patients undergoing PVE with BMSC application was compared with patients undergoing PVE without BMSC application by Student's t-test.


Characterization of CD133+ Cell Preparations

FACS analysis of native BM and the positive cell fraction after selection for CD133 (Table 1) are exemplified in detail for patient 3 in Figure 2. The calculated numbers of total intrahepatically administered CD133+ cells for patients 1, 2, and 3 were 8.8, 12.3, and 2.4 × 10E6 cells (Table 1).

Figure Figure 2..

Cytofluorimetric analyses of applied progenitor cells and gain of liver volume. Aliquots of unselected bone marrow (left panels) harvested from patients and the readministered positive fractions after enrichment for stem cell marker CD133 (right panels) were analyzed by fluorescence-activated cell sorting. Upper panels demonstrate the concentration of leukocytes detected by anti-CD45 antibodies. DNA-stain propidium iodide was used to assess rate of cell viability (middle panels). Using anti-CD133 and anti-CD34 antibodies (Miltenyi Biotec), the concentration for CD133+ cells was evaluated (lower panels). The characterization of samples derived from patient 3 is exemplified.

Gain of Left Lateral Liver Volume

Examples of hepatic CT scans are displayed for patient 2 before and after intervention in Figure 3A. CT scan–evaluated projected daily hepatic growth rates for patients 1, 2, and 3 were 10.6, 11.1, and 7.9 ml per day, resulting in a gain of the left lateral liver lobe of 51%, 122%, and 43% of the preinterventional left lateral liver volume after 22, 21, and 14 days after PVE plus SC treatment, respectively (Table 1). Determination of gain in volume of segments II/III in the reference group was based on a CT-volumetric evaluation pre-PVE contrasted with volumetric data gathered 22, 23, and 26 days after PVE, respectively (data not shown). The daily mean gain of the PVE plus BMSC application group with a mean gain of 9.87 ± 1.72 standard deviation (SD) in volume of hepatic segments II/III ranged well superior to the reference group, with a mean gain of 4.03 ml per day ± 0.47 SD (p < .01), as did the relative gain in percent of TLV (0.58 ± 0.118 SD versus 0.229 ± 0.019 SD; p < .01) (Fig. 3B).

Figure Figure 3..

Daily growth rates of hepatic segments II/III projected from computed tomography (CT) volumetry. (A): Transverse slices of CT scans obtained before (day 0) and 3 weeks after (day 21) PVE of hepatic segments I and IV to VIII plus intrahepatic application of CD133+ bone marrow cells to left lateral segments are shown as an example for patient 2. Black arrows indicate the tumor. White outline in CT scans indicates left lateral liver lobe. (B): Contrasting the mean of all three bone marrow SC patients (black columns) with that of three patients receiving the same protocol, including PVE, to expand the future liver remnant volume before extensive liver surgery except the application of bone marrow SCs in the same time period (white columns). Data are presented as mean ± standard deviation. **p < .01; Student's t-test. Abbreviations: PVE, portal venous embolization; SC, stem cell; TLV, total liver volume.

Procedure-Related Adverse Events

There were no complications or reactions observed in the course of BM acquisition, PVE, and SC application. Wound infections developed in patients 1 and 2. Fever, observed for patient three 4 days after PVE, and SC application completely resolved 48 hours after initiation of an antibiotic therapy. Peak increase of serum transaminase levels after intervention was between 0 and 76 U/l. Serum bilirubin levels and coagulation parameters, like pro-thrombin time, activated partial thromboplastin time, and international normalized ratio (INR), remained normal.

Extended Hepatectomy Subsequent to PVE and Stem Cell Application

After recognition of a sufficient FLRV of segments II/III, right trisegmentectomy of the liver (segments I and IV to VIII) was performed in all three CD133+ cell–treated patients, with Rouxen-Y and bile duct reconstruction required for patients 1 and 2. Patient 2 revealed intraoperatively a primary neuroendocrine lesion of the appendix that was surgically treated by performing an ileocolic resection. A 1.0 × 0.5-cm tumor lesion of liver segment II was resected nonanatomically in this patient. Resection margins were free of tumor in all cases.

Patients were routinely kept in the intensive care unit for 2 days. Postoperative liver function demonstrated a mild insufficiency during the first days after surgery, assessed by total serum bilirubin (maximum, 2.7–6.6 mg/dl) and coagulation status (maximum INR, 1.8–1.9), mainly resolving by postoperative day 14 (Fig. 4). Markers of hepatocellular damage demonstrated a peak on days 2 and 3 (maximum aspartate amino transferase, 179–380 U/l; maximum amino alanine transferase, 272–466 U/l), quickly declining to almost reference levels by the second week postoperatively. Patient 1 developed a bile leakage requiring open reintervention. Patient 2 demonstrated a chylascos, spontaneously ceasing subsequent to a period of total parenteral nutrition.

Figure Figure 4..

Clinical chemistry before and early after extended right hepatectomy of patients receiving portal venous embolization (PVE) plus bone marrow stem cells (SCT) or PVE only. Upper panels: routinely assessed markers of liver metabolism (total serum bilirubin, black circles) and coagulation (INR; white circles), the latter representative for hepatic syntheses. Lower panels: aspartate amino transferase (AST; white circles) and amino alanine transferase (ALT; black circles) as markers of hepatocellular damage were evaluated. Abbreviation: INR, international normalized ratio.

Of non-CD133+–treated patients, 1 and 3 were never resected, due to development of tumor mass in the left lateral segment. On control patient 2, trisegmentectomy right plus partial pancreato-duodenectomy was performed, the latter to warrant resection margins to be tumor-free. The patient was depending on high positive end-expiratory pressure (PEEP) levels in the early course after surgery due to poor pulmonary performance. A portal vein thrombosis, most likely attributable to a PEEP-triggered compromised hepatic outflow, was resolved by revision of the portal vein. However, multiorgan failure led to death on the 10th postoperative day. Liver function markers for this patient are shown in Figure 4.


Our attempt presented here to shorten the time to sufficient proliferation of the FLRV is based on the hypotheses that autologous CD133+ BM cells, when administered to the liver, may mimic and therefore augment physiological hepatic regeneration [10]. CD133 seemed promising as a selection marker for BMSCs, because a significant increase of peripheral CD133+/CD14+ leukocytes was observed after partial loss of liver tissue subsequent to hepatectomy, not observed for other major abdominal surgery. These CD133+ cells demonstrated in the same study a capacity to differentiate in vitro into a hepatic lineage [29].

The mechanism for homing of cells to the liver remains speculative. However, the damaged liver is known to express chemokines and possible chemotactants, such as stroma-derived factor 1 and others, discussed to participate in the concert of SC homing from extrahepatic sources to the liver [30]. Hepatic engraftment from extrahepatic progenitor cells is accelerated in cases of liver damage if contrasted with noninjured liver tissue [31]. To improve hepatic homing to the left-lateral liver lobe in our approach, CD133+ progenitor cells were applied after PVE of contralateral liver segments, the latter representing a strong stimulus for liver proliferation of nonembolized hepatic segments.

CT scan volumetry was used as a reliable tool to monitor TLV and FLRV [9, 32]. Mean daily hepatic growth rates were 2.5-fold higher if contrasted with a comparable group of three patients that had been subjected to PVE without BMSC application. Daily growth rates were also substantially higher than the means of a previously reported patient group that was also treated with the same therapeutic concept without application of CD133+ BMSCs [9]. These data are nonrandomized and hence may not directly be comparable; however, our early experience suggests that CD133+ cells have the therapeutic potential to augment liver regeneration processes when applied to the proliferating liver. Despite the small number of treated patients, it is interesting to note that absolute numbers of applied CD133+ cells went parallel to the extent of left-lateral growth rates.

The therapeutic potential of hematopoietic SCs to support liver regeneration subsequent to hepatic injury is still fairly unexplored. In a recent report, appearance of donor-derived cells in CCl4-mediated acutely damaged regenerating livers of female rats —that were subjected to systemic application of male, unselected BM cells—was demonstrated [33].

The mechanisms by which extrahepatic hematopoietic SCs may repopulate the regenerating liver is still under discussion. Conversion to liver cells via cell fusion [34, 35] or as transdifferentiation without fusion [19, 36] may occur. BMSCs as source for intrahepatic oval cells are hypothesized as another way to support liver regeneration [10]. Oval cells [37] are assumed to act as intrahepatic SCs [38, 39] with the capacity to differentiate both into hepatocytes [40] and bile duct cells [41]. Especially in scenarios of higher level of hepatocytic damage, this pathway may play a significant role [42, 43]. A possible contribution of low numbers of BMSCs to liver regeneration as therapeutically applied here could be explained by high concentrations of CD133+ cells that were used. In addition, direct portal application may have eased to some extent the homing mechanisms, physiologically required to attract circulating extrahepatic SCs to the regenerating liver. In a preclinical model, it has been reported that BMSCs, if applied to the portal vein, demonstrate a high percentage of first-pass entrapment in the liver [44]. However, the mechanism by which CD133+ BMSCs could accelerate liver proliferation in our therapeutic approach remains to be further investigated.

Despite the small number of patients and the lack of an adequately sized randomized control group in this study, data from the patients studied suggest that this novel therapeutic approach using the application of CD133+ cells to the portal vein may bear the potential for augmentation of liver regeneration before extensivehepatectomyinaclinicalsetting.Acontrolledtrialisinitiated to settle this issue of effectiveness in a larger number of patients.


J.S.a.E. and W.T.K contributed equally to this manuscript.