Early cytokine responses after percutaneous magnetic resonance imaging guided laser thermoablation of malignant liver tumors

www.elsevier.com/locate/issn/10434666 Cytokine 34 (2006) 278–283

Early cytokine responses after percutaneous magnetic resonance imaging guided laser thermoablation of malignant liver tumors Raija Kallio a

a,*

, Roberto Sequeiros b, Helja¨-Marja Surcel c, Pasi Ohtonen Heikki Kiviniemi e, Hannu Syrja¨la¨ f

d,e

,

Department of Oncology and Radiotherapy, Oulu University Hospital, P.O. Box 22, FIN-90029 OYS, Oulu, Finland b Department of Radiology, Oulu University Hospital, Oulu, Finland c National Public Health Institute, Oulu, Finland d Department of Anesthesiology, Oulu University Hospital, Oulu, Finland e Department of Surgery, Oulu University Hospital, Oulu, Finland f Department of Infection Control, Oulu University Hospital, Oulu, Finland Received 13 February 2006; received in revised form 25 April 2006; accepted 4 June 2006

Abstract An early systemic response induced by magnetic resonance imaging (MRI)-guided interstitial percutaneous laser thermoablation was analyzed in 13 consecutive patients with malignant liver tumors by serum interleukin (IL)-1b, IL-6, IL-10, tumor necrosis factor (TNF)a, its receptor TNFRI, and C-reactive protein (CRP) levels up to 72 h after the procedures. Only IL-6 (p = 0.033) and TNFRI (p < 0.001) increased statistically significantly after ablation, while the TNF-a, IL-1 b, and IL-10 levels remained unchanged. The peak median CRP response was 92 mg/l. There was a dose-dependent correlation between the energy used and the maximum CRP values (s = 0.68, p = 0.013). MRI-guided laser thermoablation induced an early systemic inflammatory reaction with statistically significantly elevated IL-6, TNFRI, and CRP levels but not TNF-a or IL-10 levels and without procedure-related complications, favoring this procedure as a safe therapeutic alternative for well-selected patients with liver tumors.  2006 Elsevier Ltd. All rights reserved. Keywords: Acute phase response; C-reactive protein; Interleukin-6; Receptor of tumor necrosis factor; Hepatic malignancy; Thermoablation

1. Introduction According to the most recent reports, interstitial laser thermotherapy is minimally invasive, palliative, and even curative in the treatment of patients with primary or secondary hepatic tumors [1–10]. It has also been suggested that percutaneous rather than laparotomy thermoablation methods would be preferable [9]. Of the ablative methods, magnetic resonance imaging (MRI)-guided laser ablation

*

Corresponding author. Fax: +358 8 3152452. E-mail addresses: raija.kallio@oulu.fi (R. Kallio), roberto.blanco@ oulu.fi (R. Sequeiros), helja-marja.surcel@ktl.fi (H.-M. Surcel), pasi. ohtonen@oulu.fi (P. Ohtonen), heikki.kivineimi@ppshp.fi (H. Kiviniemi), hannu.syrjala@ppshp.fi (H. Syrja¨la¨). 1043-4666/$ - see front matter  2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.cyto.2006.06.003

of the liver allows the most accurate differentiation between malignant and healthy tissues [2]. In the early phase of an inflammatory process, such as tissue injury, pro-inflammatory cytokines are released by the activation of macrophages and monocytes at the inflammatory site [11]. Pro-inflammatory cytokines, i.e., tumor necrosis factor-a (TNF-a), interleukin (IL)-lb, and IL-6, have both local and systemic effects. IL-6, in association with IL-1b and TNF-a, induces the production of acute phase proteins, including C-reactive protein (CRP), in hepatocytes, which are the major source of these proteins [12,13]. TNF-a is a pleiotropic inflammatory cytokine, which induces neutrophil proliferation during inflammation, but also induces neutrophil apoptosis upon binding to its receptor (TNFR) [14,15]. Soluble TNF-R is hypothesized to decrease the responsiveness of TNF-a by

R. Kallio et al. / Cytokine 34 (2006) 278–283

binding to soluble or membrane-bound TNF-a [13]. The activity of pro-inflammatory cytokines is also counteracted by the production of other anti-inflammatory cytokines, including IL-10 [14,15], which, for example, suppresses TNF-a release by macrophages [13]. There is some knowledge about the cytokine changes due to tissue injury in association with different liver ablation procedures performed during laparotomy or laparoscopy [16–20]. As far as we know, however, there are no data on the inflammatory reaction induced by percutaneous MR-guided laser ablation. We assessed the early systemic response to tissue injury after MRI-guided laser ablation measured by serum pro-inflammatory and anti-inflammatory cytokines as well as CRP levels in patients with different liver tumors. 2. Materials and methods 2.1. Patients The institutional ethical board of our hospital approved the study plan in accordance with the ethical standards of the committee responsible for human experimentation (institutional or regional) and with the Helsinki Declaration of 1975, as revised in 1983. Informed oral and written consent was acquired from every patient. Between September 2000 and May 2003, 13 consecutive patients with a malignant liver tumor but without disseminated disease and otherwise eligible for local ablative treatment were treated prospectively with interstitial MRI-guided laser ablation therapy. Cancer staging was carried out according to the TNM classification [21]. 2.2. Description of the laser ablation procedure The MRI-guided interventional procedure has been described in more detail elsewhere [22]. Briefly, a low-field open-configuration MRI scanner (Outlook Proview, 0.23T, Philips Medical Systems, Finland) with optical instrument guidance hardware and software was used. All patients received a single prophylactic dose of 1.5 g of cefuroxime intravenously before the procedure. Depending on the patient’s age, the localization of the lesion, and the patient’s co-operation, the procedures were performed under spinal anesthesia in nine cases and under general anesthesia in four cases, with local anesthetics (lidocain 5%) administered subcutaneously to the treatment area to alleviate post-procedural local pain. The laser device used was of the Nd-Yag type, maximum power 100 W (Fibertom medilas, Dornier Medizin technik, Germany). A diffusion type laser fiber (Dornier Medizin technik TM, Germany and Somatex GmbH, Germany) with a diameter of 400 lm was used with an irrigation system. When several fibers were needed, a laser beam splitter was used (Dornier Medizin technik TM, Germany). CBASS (completely balanced steady-state, True-FISP) imaging was used for lesion localization, instrument guid-

279

ance, and thermal monitoring. Ultrasound (Toshiba 8000 TM) was used as an auxiliary modality to guide the percutaneous puncture needle into the target. A MRI-compatible irrigated laser set (Somatex GmbH TM, Germany) was used to administer the treatment. The final position of the laser set was confirmed with MR imaging. The laser treatment was conducted using a constant energy flow and power of 10 W/s/cm as per active part of diffuser laser fiber. Laser fibers with active tip lengths of 2 and 3 cm were used. When a signal void reaching the tumor borders and beyond it with a 0.5 cm margin was detected, the treatment was considered complete. After the treatment, the patients were monitored for 6 h in the recovery room. The patients were from farther away and stayed at hospital for four days due to the need to take repeated blood samples over a period of 72 h. They were discharged from hospital on the fourth day after a clinical follow-up examination and control MRI imaging. 2.3. Cytokine analyzes Blood samples were collected before and after (2, 4, 6, 12, 24, 36, >48, 56, 72 h) the laser ablation. Serum was separated from each specimen within 4 h of sampling and stored at 70 C until analysis. Cytokine analysis were performed by enzyme immunoassay method (ELISA) using commercially available ELISA kits (ELISA DuoSet, R&D systems, Minneapolis, USA) for human Interleukin(IL) 1b, IL-6, IL-10, (TNF)-a and soluble TNF receptor 1 (TNF-R1) according to the manufacturer’s instructions. Cytokine concentrations of undiluted samples were determined as pg/ml by comparing the mean optical density (A450 mn) of duplicated samples to the optical density of a constructed standard curve of the corresponding cytokine. The lower detection limits for the tests were 3 pg/ml for IL-1b and IL-10 and 10 pg/ml for IL-6, TNF-a, and TNF-R1. 2.4. C-reactive protein measurement The CRP measurements were performed in our hospital laboratory using an enhanced turbidimetric test (Cobas Integra 700 TM, Hoffmann-La Roche Ltd., Switzerland). The detection limit of the assay was 3 mg/l. 2.5. Statistical analysis Summary measurements are expressed as median with range or interquartile (IQR) range from the 25th to 75th percentile. Friedman’s non-parametric test was used to evaluate the changes over time. Kendall’s rank correlation coefficient (s) was calculated between the energy used and the peak CRP values. For statistical purposes, CRP values less than 3 mg/l were replaced by value of 2 mg/l. Twotailed p-values are reported. SPSS (version 11.5) was used for analysis.

280

R. Kallio et al. / Cytokine 34 (2006) 278–283

3. Results

4. Discussion

The essential demographic and clinical data of the study group are presented in Table 1. Two of the metastatic patients would have been operable (patients no 1 and 9), but they refused surgery. Eight of the thirteen patients were males and five females. Their median age was 57 years with a range from 41 to 80 years. The median maximum diameter of the tumors was 3.2 cm with a range from 1.5 to 6.0 cm. The median energy used in the thermal ablations was 64 W (range from 64 to 192 W). The ablated liver volume was less than 20% in all cases. None of the patients had any inflammatory cascade associated complications involving end organs, such as respiratory failure or impairment of renal function or the coagulation system. Moreover, no procedure-related complications were observed. Complete necrosis was observed in 12 out of 13 patients one month after the procedure, and none of the patients had local recurrence after 12 months. The 3-year survival of these patients was 53.8% after the thermoablations. Of the five measured serum cytokines: TNF-a, IL-1b, IL-6, IL-10, and TNFRI, only IL-6 (p = 0.033) and TNFRI (p < 0.001) increased statistically significantly after the thermal ablation procedure (Fig. 1). There was no correlation between the maximum levels of any measured cytokines and the amount of energy used during the 56-h follow-up. There was no correlation between the maximum temperature after the procedure and the peak CRP or the TNF-a, IL-1b, IL-6, IL-10, and TNFRI levels (data not shown). The peak median CRP value was 92 mg/l with IQR from 30 to 119 mg/l. The CRP values clearly increased for up to 72 h (p < 0.001, Fig. 1). Moreover, there was a statistically significant correlation between the energy used and the maximum CRP values (s = 0.68, p = 0.013) as well as between tumor size and the peak CRP value after the procedure (s = 0.51, p = 0.017).

Our results show that percutaneous MRI-guided laser thermoablation of different hepatic tumors induces a systemic responses reflected as a statistically significant increase of IL-6, TNFRI, and CRP, but not in IL-1b, TNF-a, and IL-10 without procedure related complications or end-organ dysfunction as a consequence of systemic inflammation. As far as we known, there are only two earlier human studies concerning cytokine patterns in association with local ablation therapy of liver tumours [16,17]. In these studies, the samples were small and heterogeneous, and the ablative procedures were performed during laparotomy (see Table 2). In a third study, 16 hepatocellular cancer patients were treated with either open or laparoscopic microwave coagulation therapy [18]. In addition, there are two experimental studies on cytokine patterns in liver ablation [19,20]. In one human study on cryotherapy, both TNF-a and IL6 increased significantly after the procedure [16, Table 2]. Similarly, in one experimental study, cryotherapy caused a more profuse increase of TNF-a and IL-1b than did radiofrequency ablation (RFA) [20, Table 2], whereas in another experimental study, electrolytic liver ablation did not induce an increase of TNF-a or IL-1b [19, Table 2]. In another human study, more than 30% of the liver mass was ablated by radiofrequency therapy in 13 out of 17 patients [17, Table 2]. Despite the laparotomy and the extensive ablation procedures, serum TNF-a, IL-1b, and IL-6 concentrations did not increase. Among 16 hepatocellular cancer patients, microwave coagulation via either laparoscopy or laparotomy induced a clear increase of IL-6 concentration on the first postoperative day [18]. In our series of comparable size (n = 13), IL-6 concentrations also increased significantly after MRI-guided percutaneous thermoablation, whereas TNF-a and IL-1b did not increase. The variation

Table 1 Demographic data, type, and stage of underlying cancer, C-reactive protein (CRP) level (mg/l, before and maximum value after procedure), energy (W) used, and survival time of 13 consecutive patients with hepatic malignancy treated with thermoablation Patient no.

Sex/age (years)

Cancer type/stage

Size of tumors (cm)

Segment location of tumors

Number of ablation procedures

Peak temperature (C) within 48 h

CRP before (mg/l)

Peak CRP after the procedure (mg/l)

Used energy

Survival after procedure (months)

1 2 3 4 5 6 7 8 9 10 11 12 13

M/53 M/57 M/74 F/41 F/54 M/80 M/65 F/65 F/55 M/69 M/43 F/75 M/55

Rectal/IV Carcinoid/IV Sigmoid/IV Breast/IV Pulmonary/IV HCC/II HCC/II Ovarian/IV Rectal/IV Rectal/IV Z–E/IV HCC/II Prostatic/IV

3.8 2.3 6.0 2.1 5.5 3.5 3.0 3.0 1.5 3.5 2.7 2.7 4.2

5,6,8 2,3 4 6,7 7 8 6,8 7 7 5/6? 8 3,4 8

6 2 3 2 6 2 2 3 2 5 2 2 3

38.0 37.5 38.5 38.0 37.8 37.9 38.0 38.5 37.6 36.6 37.2 38.2 39.0

11 <3 23 5 27 2 17 — <3 5 <3 <3 5

235 9 234 25 92 141 132 163 89 95 15 34 74

192 64 96 64 192 64 64 96 64 160 64 64 96

39a 43a 61 65a 6 44a 45a 5 5 8 42a 6 9

a

Alive in January 2006. HCC, hepatocellular cancer; Z–E, Zollinger–Ellison syndrome.

R. Kallio et al. / Cytokine 34 (2006) 278–283 500

281

1000 900

TNF-α 400

IL-10

800 700 600

pg/ml

pg/ml

300

200

500 400 300

100

200 100

0

0 0

10

20

30

40

50

60

70

0

30

10

20

40

50

60

70

40

50

60

70

40

50

60

70

3000

IL-1β

TNFRI

25

2500

20

p < 0.001

2000

15

pg/ml

pg/ml

30

10

1500

1000

5

500

0 0 0

10

20

30

40

50

60

70

0

200

10

20

30

200

IL-6 p = 0.033

180

CRP

160

p < 0.001

150 140

100

mg/L

pg/ml

120 100 80 60 50 40 20 0

0 0

10

20

30

40

Time (hours)

50

60

70

0

10

20

30

Time (hours)

Fig. 1. Behavior of the serum levels of tumor necrosis factor a (TNF-a), interleukin-1b (IL-1b), IL-6, IL-10, TNFRI, and CRP (medians and 25 and 75% percentiles) before and after MRI-guided laser ablation for up to three days in 13 patients. p-value indicates the change over time (Friedman test).

282

R. Kallio et al. / Cytokine 34 (2006) 278–283

Table 2 The changes of serum cytokine levels, i.e., tumour necrosis factor alpha (TNF-a), interleukin-1 beta (IL-1b), IL-6, and receptor I of tumor necrosis factor (TNFRI), from the baseline lane after hepatic ablative procedures in earlier human and experimental studies—laparotomy (surgery) and (Ref. no.)—and the present study Method of hepatic intervention

Study population

No. of subjects

TNF-a

IL-1b

IL-6

TNFRI

Surgery +/ , liver resection +/ cryothreapy (16): Surgery + cryotherapy only (16): Surgery+ radiofrequency ablation (17)

Hepatic metastases and 3 patients with hepatocellular cancer Hepatic metastases and 3 patients with hepatocellular cancer Hepatocellular cancer Hepatocellar cancer Pig Swine Swine Hepatic metastases and 3 hepatocellular cancer

45 5 17

› Ma

ND M

› M

ND M

9 16 20 5 5 13

ND ND M › ›› M

ND ND M › ›› M

› › ND ND ND ›

M M ND ND

Surgery + microwave coagulation (18) Laparoscopy + microwave coagulation (18) Surgery + electrolytic ablation (19) Surgery + radiofrequency ablation (20) Surgery + cryotherapy (20) Percutaneous MRI-guided laser thermoablation (present study) a

›

M, no change; ND, not done; ›, statistically significant increase.

in the changes of the studied cytokine levels may be due to the different type of procedure (open surgery and different type of ablations) and the heterogeneity of the study populations (different types of cancer, human versus experimental populations). A dose-dependent rise in the TNF-a and IL-6 concentrations after hepatic cryotherapy has been reported earlier [16]. In our study with MRI-guided percutaneous thermoablation, no such correlations were seen between the studied cytokines and the dose of energy used. This can be explained by the procedure itself, but also by the fact that the ablated tumors were relatively small. We showed a significant increase of TNFRI (p55) production after ablation. Soluble TNF-R is hypothesized to decrease the responsiveness of TNF-a by binding to soluble or membrane-bound TNF-a [13], and this might explain why TNF-a did not increase after thermoablation in our series. In an earlier study with RFA during laparotomy, TNFRI concentrations did not increase [17]. The same was true of another series of hepatocellular cancer patients [18]. In one earlier study [17] and in the present study, IL10 concentrations did not increase. The severity of tissue destruction reflects on the serum CRP concentration in surgery, for example [23]: the larger the operation, the higher the postoperative CRP peak. As far as we are aware, there is only one experimental study of local ablative therapy of the liver and CRP behavior where a local electrolytic liver ablation procedure during laparotomy induced a transient and dose-dependent elevation of CRP [19]; in the sham operation group, CRP reached a peak at 24 h but decreased thereafter, while in the group with laparotomy and electrolytic liver ablation, CRP increased for up to 72 h. In a clinical study with hepatocellular cancer patients, a statistically significant increase of postoperative CRP values was observed after both open and laparoscopic microwave coagulation therapy [18]. In our prospective series of patients with percutaneous MRI-guided thermal laser ablation, we also saw a dose-related CRP rise up to 72 h.

The less invasive local ablative methods are increasingly employed as a part of multimodality therapy for primary and secondary malignant liver tumors. MRI-guided laser ablation of the liver is a well controlled, safe and accurate procedure for the coagulation of metastatic tissue minimizing the destruction of healthy tissues [2,24]. This might be advantageous in the treatment of primary and secondary hepatic tumors in comparison to major surgical procedures [25–27]. In our series of 13 patients, MRI-guided percutaneous laser thermoablation did not seem to cause any systemic inflammatory response syndrome associated organ dysfunctions or procedure-related complications, encouraging the use of this approach for the treatment of liver tumors even repeatedly when needed. MRI-guided percutaneous laser thermoablation, without any procedure-related complications, induced an early systemic inflammatory reaction with statistically significant increase of the IL-6, TNFRI, and CRP levels but not in TNF-a and IL-10 levels. According to our results this procedure seems to be a good alternative for therapeutic approach in well-selected patients with primary or secondary liver tumours. References [1] Yan DB, Clingan P, Morris DL. Hepatic cryotherapy and regional chemotherapy with or without resection for liver metastases from colorectal carcinoma. Cancer 2003;98:320–30. [2] Erce C, Parks RW. Interstitial ablative techniques for hepatic tumors. Br J Surg 2003;90:272–89. [3] Nikfarjam M, Christophi C. Interstitial laser thermotherapy for liver tumors. Br J Surg 2003;90:1033–47. [4] Livraghi T, Solbiati L, Meloni F, Ierace T, Goldberg SN, Gazelle GS. Percutaneous radiofrequency ablation of liver metastases in potential candidates for resection: the ‘‘test-of-time approach’’. Cancer 2003;97:3027–35. [5] Nagashima I, Takada T, Matsuda K, Adachi M, Nagawa H, Muto T, et al. A new scoring system to classify patients with colorectal liver metastases: proposal of criteria to select candidates for hepatic resection. J Hepatobiliary Pancreat Surg 2004;11:79–83. [6] Elias D, Baton O, Sideris L, Boige V, Malka D, Liberale G, et al. Hepatectomy plus intraoperative radiofrequency ablation and che-

R. Kallio et al. / Cytokine 34 (2006) 278–283

[7]

[8]

[9]

[10]

[11] [12] [13]

[14]

[15]

[16]

[17]

motherapy to treat technically unresectable multiple colorectal liver metastases. J Surg Oncol 2005;90:36–42. Hong SN, Lee S-Y, Choi MS, Lee JH, Koh KC, Paik SW, et al. Comparing the outcomes of radiofrequency ablation and surgery in patients with a single small hepatocellular carcinoma and wellpreserved hepatic function. J Clin Gastroenterol 2005;39:247–52. Stelzner S, Hellmich G, Koch R, Ludwig K. Factors predicting survival in stage IV colorectal carcinoma patients after palliative treatment. A multivariate analysis. J Surg Oncol 2005;89:211–7. Jungraithmayr W, Burger D, Olschewski M, Eggstain S. Cryoablation of malignant liver tumors: results of a single center study. Hepatobiliary Pancreat Dis Int 2005;4:554–60. Gaffke G, Gebayer B, Knollmann FD, Helmberger T, Ricke J, Oettle H, et al. Use of semiflexible applicators for radiofrequency ablation of liver tumors. Cardiovasc Interventional Radiol 2006;29:270–5. Gabay C, Kushner I. Acute-phase proteins and other systemic responses to inflammation. N Engl J Med 1999;340:448–54. Pepys MB, Hirschfield GM. C-reactive protein: a critical update. J Clin Invest 2003;111:1805–12. Tzianabos AO, Wetzler LM. Cellular communication. In: Pier GB, Lyczak JB, Wetzler LM, editors. Immunology, infection, and immunity. Washington, (DC): ASM Press; 2004. p. 343–69. Carswell E, Old L, Kassel R, Green N, Fiore N, Williamson B. An endotoxin-induced serum factor that causes necrosis of tumors. Proceedings of the National Academy of Science 1975;72:3666–70. Murray J, Barbara J, Dunkley S, Lopez A, Van Ostade X, Condliffe I, Haslett C, Chilvers E. Regulation of neutrophil apoptosis by tumor necrosis factor-alpha: requirements for TNF-R55 and TNF-R75 for induction of apotosis in vitro. Blood 1997;90:2772–83. Seifert JK, Stewart GJ, Hewitt PM, Bolton E, Junginger T, Morris DL. Interleukin-6 and Tumour Necrosis Factor-a levels following hepatic cryotherapy: association with volume and duration of freezing. World J Surg 1999;23:1019–26. Schell SR, Wessels FJ, Abouhamze A, Moldawer LL, Copeland EM. Pro- and antiinflammatory cytokine production after radiofrequency ablation of unresectable hepatic tumors. J Am Coll Surg 2002;195:774–81.

283

[18] Sadamori H, Yagi T, Kanaoka Y, Morimoto Y, Inagaki M, Ishikawa T, et al. The analysis of the usefulness of laparoscopic microwave coagulation therapy for hepatocellular carcinoma in patients with poor hepatic reserve by serial measurements of IL-6, cytokine antagonists, and C-reactive protein. Surg Endosc 2003;17:510–4. [19] Teague BD, Court FG, Morrison CP, Kho M, Wemyss-Holden SA, Maddern GJ. Electrolytic liver ablation is not associated with evidence of a systemic inflammatory response syndrome. Br J Surg 2004;91:178–83. [20] Ng KK, Lam CM, Poon RT, Shek TW, To JY, Wo YH, et al. Comparison of systemic responses of radiofrequency ablation, cryotherapy and surgical resection in porcine liver model. Ann Surg Oncol 2004;11:650–7. [21] Sobin LH, Wittekind Ch, editors. TNM classification of malignant tumours. New York: Wiley–Liss; 2002. [22] Sequeiros RB, Hyvonen P, Sequeiros AB, Jyrkinen L, Ojala R, Klemola X, Vaara T, Tervonen O. MR imaging-guided of osteoid osteomas with the use of optical instrument guidance at 0.23T. Eur Radiol 2003;13:2309–14. [23] Schietroma M, Carlei F, Farnchi L, Mazzotta C, Sozio A, Lygidkis NJ, et al. A comparison of serum interleukin-6 concentration in patients treated by cholecystectomy via laparotomy or laparoscopy. Hepatogastroenterology 2004;51:1595–9. [24] Vogl TJ, Straub R, Eichler K, Sollner O, Mack MG. Colorectal carcinoma metastases in liver: laser-induced interstitial thermotherapy-local tumor control rate and survival data. Radiology 2004;230:450–8. [25] Colacchio TA, Yeager MP, Hildebrand LW. Perioperative immunomodulation in cancer surgery. Am J Surg 1994;167:174–9. [26] Isbert C, Boerner A, Ritzx J-P, Schuppan D, Buhr HJ, Germer CT. In situ ablation of experimental liver metastases delays and reduces residual intrahepatic tumour growth and peritoneal tumour spread compared with hepatic resection. Br J Surg 2002;89:1252–9. [27] Koch M, Kienle P, Hinz U, Antolovic D, Schmidt J, Herfarth C, et al. Detection of hematogenous tumor cell dissemination predicts tumor relapse in patients undergoing surgical resection of colorectal liver metastases. Ann Surg 2005;241:199–205.