- Email: [email protected]
Pro- and antiinflammatory cytokine production after radiofrequency ablation of unresectable hepatic tumors
Pro- and Antiinflammatory Cytokine Production after Radiofrequency Ablation of Unresectable Hepatic Tumors Scott R Schell, MD, PhD, FACS, Frank J Wessels, MD, Amer Abouhamze, BS, Lyle L Moldawer, PhD, Edward M Copeland III MD, FACS Experience using radiofrequency ablation (RFA) for treating unresectable hepatic malignancies is expanding, with promising outcomes and fewer complications compared with cryotherapy. This study examined systemic inflammatory responses after RFA as measured by the appearance of postoperative symptoms and cytokine production. STUDY DESIGN: Seventeen patients (11 men, 6 women) aged 40 to 85 years (mean 64.2 years) with unresectable primary and metastatic hepatic tumors underwent RFA. Mean liver volume treated with RFA was 35.3% ⫾ 3.6% (SEM) (median 36.8%). Plasma cytokines (tumor necrosis factor-␣, interleukin [IL]-1, IL-1ra, IL-6, IL-8, IL-10, p55, and p75) were measured from anesthesia induction through 48 hours after RFA. Ex vivo whole-blood cytokine production was measured at baseline, 24 hours, and 48 hours after RFA. RESULTS: Cytokine and cytokine-receptor production were not notably altered by RFA. Ex vivo wholeblood endotoxin stimulation indicated that intrinsic cellular immune function remained intact after treatment, although modest decreases in stimulated tumor necrosis factor ␣ production were observed 24 to 48 hours after RFA. Variceal bleeding, hepatic failure, and death occurred in one patient 30 days after RFA. None of the remaining patients exhibited tachycardia or hypotension. Fevers (ⱖ38.5°C) developed in three patients during the first 48 hours postoperatively. There was no association between plasma cytokines and postoperative complications. CONCLUSIONS: In contrast to previous reports using cryotherapy, systemic inflammatory responses as measured by increased cytokines were not observed after RFA. The cryotherapy-induced “cryoshock” phenomenon was not observed in patients undergoing RFA in our study. We conclude that RFA ablation is fundamentally different than cryotherapy and apparently does not stimulate Kupffer and other hepatic macrophages to produce proinflammatory cytokines. ( J Am Coll Surg 2002; 195:774–781. © 2002 by the American College of Surgeons) BACKGROUND:
Radiofrequency ablation (RFA) is gaining increasing approval in the treatment of unresectable primary and metastatic hepatic tumors.1-4 Hepatic resection remains the standard of care for resectable lesions.5-7 Ongoing international experience with the use of RFA for unresectable tumors has clearly demonstrated its safety over cryotherapy. Initial encouraging reports of the efficacy of RFA have prompted many users to adopt this procedure over
cryoablation for these unresectable hepatic tumors. With introduction of progressively larger RFA arrays and capability to generate larger ablations, combined with the applicability of minimally invasive surgery to RFA, the relative advantages previously held by cryotherapy appear to be waning. In addition, cases of major systemic inflammatory complications, including hepatic and renal failure, coagulopathy, and syndromes similar to adult respiratory distress syndrome (ARDS), are observed in patients treated with cryotherapy, particularly in patients who undergo ablation of large volumes of liver (⬎30% of total hepatic volume).8-11 Appearance of “cryoshock”-like clinical entities12 has not been reported in patients treated with RFA, and this absence of marked inflammatory response has remained
No competing interests declared.
Received January 18, 2002; Revised April 22, 2002; Accepted June 25, 2002. From the Departments of Surgery and Molecular Genetics and Microbiology, University of Florida College of Medicine, Gainesville, FL; and Surgical Service, Malcolm Randall Veterans Affairs Medical Center, Gainesville, FL. Correspondence address: Scott R Schell, MD, PhD, FACS, Department of Surgery, University of Florida College of Medicine, 1600 SW Archer Rd, Box 100286, Gainesville FL 32610-0286.
© 2002 by the American College of Surgeons Published by Elsevier Science Inc.
774
ISSN 1072-7515/02/$21.00 PII S1072-7515(02)01333-9
Schell et al
Vol. 195, No. 6, December 2002
Cytokine Release after Radiofrequency Ablation
775
Table 1. Patient Demographics Patient Age no. (y) Gender
Diagnosis
1 2 3 4 5
50 67 66 85 72
F M M M M
CRC CRC HCC with marked cirrhosis CRC HCC with marked cirrhosis
6 7 8 9 10 11 12 13 14 15 16 17
40 73 73 63 51 58 63 69 48 77 73 63
F M M M F M M F F F M M
Carcinoid CC CRC GI stromal tumor Leiomyosarcoma CRC HCC CRC BC CRC CC CRC
Length of stay (d)
Liver volume ablated*
2 3 2 7 30
14.7% 36.8% 35.8% 30.7% 19.8%
3 3 4 6 5 3 4 3 5 5 3 2
37.9% 37.3% 9.7% 18.8% 55.3% 35.8% 44.9% 31.5% 46.5% 48.6% 58.3% 48.1%
Complications
None None None Ileus, urinary tract infection Upper GI hemorrhage, bacteremia, hepatic failure, death on postoperative day 30 None None None None, underwent transfusion for low preoperative hematocrit None None None None None None None, underwent transfusion for low preoperative hematocrit None
*As percentage of calculated liver mass. BC, breast carcinoma; CC, cholangiosarcoma; CRC, colorectal carcinoma metastases; GI, gastrointestinal; HCC, hepatocellular carcinoma.
intriguing from the perspective of how RFA affects innate immunity and production of pro- and antiinflammatory cytokines after treatment. Previous reports have examined elaboration of proinflammatory cytokines after a variety of surgical procedures, ranging from minimally invasive surgery13-16 to major vascular surgical reconstruction.17 In these studies, elevation of proinflammatory cytokines in the patients’ plasma was directly associated with major systemic responses. No previously published reports have evaluated production of pro- or antiinflammatory cytokines in patients undergoing RFA treatment. Several authors have published reports using rat models of RFA and cryotherapy and have examined cytokine production, lung inflammation, and the histopathology of lung injury after treatment.18-21 These studies demonstrated a direct association between cryotherapyinduced lung injury and NF-B activation and production of NF-B–dependent cytokines, tumor necrosis factor (TNF)-␣, and macrophage inflammatory protein-2. The purpose of this study was to examine production of pro- and antiinflammatory cytokines in patients undergoing RFA treatment for unresectable primary and metastatic hepatic tumors.
METHODS Patient recruitment
Patients undergoing RFA were recruited from Shands Hospital at the University of Florida, Gainesville, FL. The Institutional Review Board at our institution approved experimental protocols, and signed informed consent was obtained from all patients before enrollment. The enrollment and study of research patients at our institution are in strict compliance with the Helsinki Declaration of 1976. Seventeen patients undergoing RFA for unresectable primary or metastatic liver tumors agreed to blood sampling before and for 48 hours after RFA treatment for measurement of pro- and antiinflammatory cytokines. Patient demographics
Seventeen patients (11 men, 6 women) with unresectable primary (hepatocellular carcinoma, three patients) or metastatic (colorectal carcinoma, eight patients; cholangiocarcinoma, two patients; sarcoma, two patients; breast, one patient; carcinoid, one patient) hepatic tumors were treated in this study. Patient age ranged from 40 to 85 years (mean⫾SEM, 64.2⫾2.8 years). Patient demographics, diagnoses, and postoperative length of stay are described in Table 1.
776
Schell et al
Cytokine Release after Radiofrequency Ablation
Study protocol
Preoperative imaging using either CT or MRI was obtained for all patients before they underwent laparotomy and abdominal exploration. After the presence of extrahepatic metastases was excluded, the size, location, geometry, and histology of each hepatic lesion were confirmed using intraoperative ultrasound and ultrasoundguided core biopsy. After biopsy confirmation of histology, each metastasis was treated using LeVeen-type monopolar RFA arrays and either the RF2000 or RF3000 RFA generator (Radiotherapeutics, Inc, Palo Alto, CA), using techniques described previously.1,4 Each lesion was treated with at least two cycles of RFA, pursuant to our standard ablation protocol, with the first RFA cycle continued until either maximal tissue impedance was obtained, or until a total of 15 treatment minutes had elapsed. After a 2-minute “cool-down” period, a second cycle of RFA was performed using the time and impedance criteria just described. Hepatic tumors that were larger than our probe size were divided into overlapping quadrants, and each quadrant was treated separately using the protocol just described. A margin of normal liver of at least 1 cm was included with each ablation, and ablated margins were verified using intraoperative ultrasonography. The estimated volume of liver treated using RFA was calculated as a percentage of total hepatic volume using methods described previously, with liver mass (in grams) calculated as (patient weight in kilograms)/70)⫻(4/ 7)⫻1,000.22 Every treated lesion was measured using intraoperative ultrasound, and volume of ablated tissue was calculated after adding a 1-cm margin of normal liver to the largest radius of each lesion. The volumes of all ablated lesions were then summed and divided into the total liver volume and expressed as a percentage of total liver volume. The weight-based methodology used for liver volume calculations has been shown to be highly accurate in autopsy and subsequent clinical series, but this methodology might tend to somewhat overestimate actual liver volumes ablated. Postoperatively, patients were admitted to the surgical oncology service and were closely monitored for fever and signs or symptoms of sepsis. Patients were encouraged to walk on the first postoperative day and were discharged home when they were able to tolerate a regular diet. After discharge, hepatic CT imaging was performed 1 week and 1 month after RFA to confirm success of treat-
J Am Coll Surg
ment and to compare effective ablated hepatic volume with that calculated during the RFA procedures. Ablated liver volume as assessed using postoperative CT imaging was observed to be highly concordant with the volumes calculated at the time of operation. Blood collection for cytokine assays and ex vivo stimulation
Blood samples (7 mL) were obtained after induction of anesthesia and immediately before commencement of RFA to provide internal baseline controls for each patient. Additional blood samples (7 mL) were obtained after completion of the final RFA treatment and thereafter at timed intervals (2 minutes and 1, 2, 4, 8, 24, and 48 hours after RFA). All blood samples were collected with dipotassium ethylenediaminetetraacetic acid, were immediately placed in ice, and were promptly centrifuged at 1,300 g at 4°C for 15 minutes; the plasma fraction was separated, aliquoted, and stored at ⫺70°C until analysis, as described previously by our laboratories.17 Additional blood samples (7 mL) were obtained for examination of ex vivo stimulated cytokine production after induction of anesthesia—to provide baseline levels for each patient—and at 24 hours and 48 hours after the final RFA treatment. Samples were collected and immediately placed in ice, and the whole blood used for ex vivo cytokine production was measured after stimulation with lipopolysaccharide, using the protocol described previously.23,24 Biochemical analyses
Freshly thawed plasma samples were assayed by ELISA for TNF␣, interleukin [IL]-1, IL-1ra, IL-6, IL-8, IL10, and the TNF shed receptors p55 and p75, as described previously by our laboratories.17 The sensitivity of the TNF-␣, IL-6, IL-8, IL-10, p55, and p75 assays varied from 32 to 128 pg/mL, as has been reported previously.17,25,26 Values obtained below the detection limits of an individual assay were treated as 0 for purposes of statistical analysis. Whole-blood samples were stimulated with LPS overnight at 37°C at final concentrations of 1 ng/mL or 1 g/mL using methods described previously.23 Whole blood was assayed by ELISA for TNF␣, IL-1, IL-1ra, IL-6, IL-8, IL-10, and TNF shed receptors p55 and p75 at 0, 24, and 48 hours after LPS stimulation using the identical assay procedures described earlier.
Vol. 195, No. 6, December 2002
Schell et al
Statistical analyses
All data are expressed as the mean ⫾ SEM. Baseline measurement data were obtained for each patient after anesthesia induction and before commencement of RFA. Within-patient changes from baseline in plasma cytokine levels were analyzed using one-way ANOVA to compare the means at all the time points, with any overall significant differences subjected to Tukey’s honestly significant difference test. Differences were considered significant if the two-sided p value was less than 0.05. RESULTS Postoperative hospital length of stay ranged from 2 to 30 days (mean 5.3⫾1.6 days). No patients developed systemic inflammatory syndromes equivalent to either cryoshock or ARDS. One patient (patient 5) had a history of marked cirrhosis and preoperative variceal bleeding. After RFA treatment, this patient experienced recurrence of variceal bleeding on the second postoperative day, which progressed to irreversible hepatic failure refractory to transjugular portosystemic shunting and systemic bacteremia. The patient died on the 30th postoperative day. Of the remaining 16 patients, none exhibited postoperative hypotension. Postoperative ileus and urinary tract infection developed in one patient (patient 4), who was discharged on the seventh postoperative day without further complications. There were no other complications. Pyrexia (ⱖ38.5°C) developed in three patients (patients 2, 5, and 8) during the first 48 hours after completion of RFA, but these temperature elevations were not associated with notable leukocytosis or hemodynamic perturbation; they resolved spontaneously with ambulation and incentive spirometry and did not require treatment with antibiotics. Postoperative hepatic CT imaging at 1 week and 1 month after RFA confirmed successful RFA treatment in all patients. When compared with baseline levels measured in each patient at induction of anesthesia, RFA treatment did not induce pronounced changes in plasma levels of the cytokines IL-1, IL-8, IL-10, or TNF-␣ (Fig. 1A). After RFA, plasma IL-1ra, IL-6, p55, and p75 concentrations were only modestly stimulated (Fig. 1B), but these levels did not reach statistical significance. There were no associations observed between the elevations in plasma cytokine appearance and the development of postoperative pyrexia occurring in patients 2, 5, and 8. RFA treatment did not alter the production of pro-
Cytokine Release after Radiofrequency Ablation
777
and antiinflammatory cytokines as measured by ex vivo LPS stimulation of whole-blood samples before and after RFA treatment. Figure 2A shows stimulated IL-1, IL-8, IL-10, and TNF-␣ production before RFA treatment and at 24 and 48 hours after treatment. Figure 2B shows LPS-stimulated IL-1ra, IL-6, p55, and p75 during the same time period. Although a modest decrease in TNF-␣ production was noted 24 to 48 hours after RFA (4.4 ng/mL versus 13.4 ng/mL control), the differences in concentration did not reach statistical significance, and no enhancement or inhibition of the other cytokines in response to LPS stimulation was observed. DISCUSSION These preliminary studies suggest that RFA does not induce a significant plasma appearance of the proinflammatory cytokines IL-1, IL-1ra, IL-6, IL-8, IL-10, TNF-␣, or the TNF shed receptors p55 and p75 in human patients undergoing RFA for unresectable primary or metastatic liver tumors. It appears that RFA does not activate production of the same proinflammatory cytokines elaborated after cryotherapy of similar liver tumors19 and does not appear to mimic the same degree of inflammatory response observed after the minimally invasive surgical procedure laparoscopic cholecystectomy.13,16 In addition, RFA does not substantially alter cytokine production after LPS stimulation of whole blood obtained from these patients. Although patients in this study were treated with RFA to achieve ablation volumes equivalent to those obtained in patients developing cryoshock after cryotherapy, our observations about these patients’ cytokine production are concordant with our institution’s observations and experience1,3,4 of patients undergoing large-volume RFA who do not exhibit symptoms consistent with the cryoshock phenomenon after ablation. With a combined total of nearly 65,000 new cases of colorectal hepatic metastases and hepatocellular carcinomas each year, and with only 25% to 35% of patients presenting with resectable lesions, there remains a considerable need for additional treatment modalities for these patients.7,27 In addition, although prospective randomized trials comparing hepatic resection and RFA for resectable lesions are still forthcoming, the observed benefits of hepatic resection—with 5-year survivals of 30% to 35% after complete resection of three to five lesions28—raise the question whether hepatic resection
778
Schell et al
Cytokine Release after Radiofrequency Ablation
J Am Coll Surg
Figure 1. Cytokine concentrations after radiofrequency ablation (RFA) treatment. (A) Interleukin (IL)-1, IL-8, IL-10, and tumor necrosis factor (TNF)-␣. (B) IL-6, IL-1ra, p55, and p75. Cytokine concentrations were assessed at time of anesthesia induction (A.I.), after RFA probe placement (RFA Start), on completion of all RFA treatments (RFA Stop), and at the time points indicated thereafter. Each data point represents the mean for the series. Error bars are ⫾ 1 SEM.
provides adequate outcomes for the risks and resources consumed. Past reports of cryotherapy, although demonstrating efficacy and versatility in treating a variety of tumor sizes, types, geometries, and intrahepatic locations,9,10,29-33 have shown that cryotherapy is also complicated by risks of coagulopathy, renal and respiratory failure, and other major
morbidities at rates, in some series, equivalent to those described for hepatic resection.8,11,12 There is increasing evidence that hepatic Kupffer cells are important participants in inflammatory responses after hepatic injury or hypoperfusion. Seifert and colleagues34 have recently demonstrated that cryoablation results in TNF-␣ and IL-6 release in a rat model. Previ-
Vol. 195, No. 6, December 2002
Schell et al
Cytokine Release after Radiofrequency Ablation
779
Figure 2. Cytokine production by LPS-stimulated whole blood after radiofrequency ablation (RFA) treatment. (A) Interleukin (IL)-1, IL-8, IL-10, and tumor necrosis factor (TNF)-␣. (B) IL-6, IL-1ra, p55, and p75. Patient whole blood was collected before RFA (0 hours) and 24 and 48 hours after completion of RFA treatment. Collected samples were treated with low (1 ng/mL) or high (1 g/mL) concentrations of lipopolysaccharide, and cytokine production was assessed. Each data point represents the mean for the series. Error bars are ⫾1 SEM.
ous studies have shown that direct endothelial damage and clinical syndromes equivalent to ARDS arise after IV infusion of TNF-␣, identifying this cytokine as the likely culprit in the previously characterized cryoshock phenomenon.12,35,36 The absence of proinflammatory cytokine production after RFA treatment does not appear related to the
body’s inability to elaborate these cytokines, because cytokine production by the patient’s endotoxin-stimulated whole blood was not observed to be notably decreased. Instead, we propose that the most likely explanation for our observations is based in the mechanism of tumor destruction achieved by using RFA, specifically, that the thermal injury created by RFA results in coagulation,
780
Schell et al
Cytokine Release after Radiofrequency Ablation
cellular protein denaturation, and subsequent cell death. Accordingly, it is reasonable to expect that adjacent Kupffer cells and their released cytokines would be subjected to the same thermal coagulation and so be rendered inactive. In contrast, cryotherapy involves freezing using liquid nitrogen–cooled probes, and cellular lysis and necrosis result from intracellular ice crystals forming as a result of the freeze-thaw cycles. Cellular necrosis, unlike apoptosis, is generally associated with an inflammatory response, particularly in organs rich in tissue macrophages. In addition, because few cytokines are inactivated by freezing, cytokines elaborated during cryoablation would remain active to induce local and systemic effects. Together, these preliminary findings suggest that hepatic tumor ablation after RFA-induced thermal coagulation has much lower systemic inflammatory potential than cryotherapy while yielding equivalent degrees of tissue ablation. The advent of RFA for the treatment of unresectable hepatic metastases has provided an additional therapeutic alternative for patients with unresectable hepatic tumors while avoiding the risks associated with cryoablation and hepatic resection. Author Contributions
Study conception and design: Schell, Wessels, Abouhamze, Moldawer, Copeland Acquisition of data: Schell, Wessels, Abouhamze, Moldawer, Copeland Analysis and interpretation of data: Schell, Wessels, Abouhamze, Moldawer, Copeland Drafting of manuscript: Schell, Wessels, Abouhamze, Moldawer, Copeland Critical revision: Schell, Wessels, Abouhamze, Moldawer, Copeland Statistical expertise: Schell, Wessels, Abouhamze, Moldawer, Copeland Supervision: Schell, Moldawer REFERENCES 1. Wessels FJ, Schell SR. Radiofrequency ablation treatment of refractory carcinoid hepatic metastases. J Surg Res 2001;95:8– 12. 2. Rose DM, Allegra DP, Bostick PJ, et al. Radiofrequency ablation: a novel primary and adjunctive ablative technique for hepatic malignancies. Am Surg 1999;65:1009–1014. 3. Siperstein AE, Rogers SJ, Hansen PD, Gitomirsky A. Laparoscopic thermal ablation of hepatic neuroendocrine tumor metastases. Surgery 1997;122:1154–1155.
J Am Coll Surg
4. Curley SA, Izzo F, Delrio P, et al. Radiofrequency ablation of unresectable primary and metastatic hepatic malignancies: results in 123 patients [see comments]. Ann Surg 1999;230:1–8. 5. Fong Y, Kemeny N, Paty P, et al. Treatment of colorectal cancer: hepatic metastasis. Semin Surg Oncol 1996;12:219–252. 6. Kavolius J, Fong Y, Blumgart LH. Surgical resection of metastatic liver tumors. Surg Oncol Clin North Am 1996;5:337– 352. 7. Fong Y, Blumgart LH. Surgical options in the treatment of hepatic metastases for colorectal cancer. Curr Probl Surg 1996; 35:336–413. 8. Sarantou T, Bilchik A, Ramming KP. Complications of hepatic cryosurgery. Semin Surg Oncol 1998;14:156–162. 9. Stewart GJ, Preketes A, Horton M, et al. Hepatic cryotherapy: double-freeze cycles achieve greater hepatocellular injury in man. Cryobiology 1995;32:215–219. 10. Haddad FF, Chapman WC, Wright JK, et al. Clinical experience with cryosurgery for advanced hepatobiliary tumors. J Surg Res 1998;75:103–108. 11. Bagia JS, Perera DS, Morris DL. Renal impairment in hepatic cryotherapy. Cryobiology 1998;36:263–267. 12. Seifert JK, Morris DL. World survey on the complications of hepatic and prostate cryotherapy. World J Surg 1999;23:113– 114. 13. Bellon JM, Manzano L, Larrad A, et al. Endocrine and immune response to injury after open and laparoscopic cholecystectomy. Int Surg 1998;83:24–27. 14. Berguer R, Bravo N, Bowyer M, Ferrick D. Measurement of intracellular gamma-interferon, interleukin-4, and interleukin-10 levels in patients following laparoscopic cholecystectomy. J Invest Surg 2000;13:161–167. 15. Helmy SA, Wahby MA, El-Nawaway M. The effect of anaesthesia and surgery on plasma cytokine production. Anaesthesia 1999;54:733–738. 16. Lemaire LC, van der Poll T, van Lanschot JJ, et al. Minimally invasive surgery induces endotoxin-tolerance in the absence of detectable endotoxemia. J Clin Immunol 1998;18:414–420. 17. Welborn MB, Oldenburg HS, Hess PJ, et al. The relationship between visceral ischemia, proinflammatory cytokines, and organ injury in patients undergoing thoracoabdominal aortic aneurysm repair. Crit Care Med 2000;28:3191–3197. 18. Washington K, Debelak JP, Gobbell C, et al. Hepatic cryoablation-induced acute lung injury. Histopathologic findings. J Surg Res 2001;95:1–7. 19. Chapman WC, Debelak JP, Wright Pinson C, et al. Hepatic cryoablation, but not radiofrequency ablation, results in lung inflammation. Ann Surg 2000;231:752–761. 20. Chapman WC, Debelak JP, Blackwell TS, et al. Hepatic cryoablation-induced acute lung injury: pulmonary hemodynamic and permeability effects in a sheep model. Arch Surg 2000;135:672–673. 21. Blackwell TS, Debelak JP, Venkatakrishnan A, et al. Acute lung injury after hepatic cryoablation: correlation with NF-kappa B activation and cytokine production. Surgery 1999;126:518– 526. 22. DeLand FH, North WA. Relationship between liver size and body size. Radiology 1968;91:1195–1198. 23. van der Poll T, Coyle SM, Moldawer LL, Lowry SF. Changes in endotoxin-induced cytokine production by whole blood after in vivo exposure of normal humans to endotoxin. J Infect Dis 1996;174:1356–1360.
Vol. 195, No. 6, December 2002
Schell et al
24. Fong YM, Marano MA, Moldawer LL, et al. The acute splanchnic and peripheral tissue metabolic response to endotoxin in humans. J Clin Invest 1990;85:1896–1904. 25. Welborn MB III, Douglas WG, Abouhamze Z, et al. Visceral ischemia-reperfusion injury promotes tumor necrosis factor (TNF) and interleukin-1 (IL-1) dependent organ injury in the mouse. Shock 1996;6:171–176. 26. Van Zee KJ, Kohno T, Fischer E, et al. Tumor necrosis factor soluble receptors circulate during experimental and clinical inflammation and can protect against excessive tumor necrosis factor alpha in vitro and in vivo. Proc Natl Acad Sci U S A 1992;89:4845–4849. 27. American Cancer Society. Cancer facts and figures. Atlanta, GA: American Cancer Society; 1998. 28. Hughes KS, Simon R, Songhorabodi S, et al. Resection of the liver for colorectal carcinoma metastases: a multiinstitutional study of patterns of recurrence. Surgery 1986; 100:278–284. 29. Adam R, Majno P, Castaing D, et al. Treatment of irresectable
30. 31. 32. 33. 34. 35. 36.
Cytokine Release after Radiofrequency Ablation
781
liver tumours by percutaneous cryosurgery. Br J Surg 1998;85: 1493–1494. Seifert JK, Morris DL. Prognostic factors after cryotherapy for hepatic metastases from colorectal cancer. Ann Surg 1998;228: 201–208. Shafir M, Shapiro R, Sung M, et al. Cryoablation of unresectable malignant liver tumors. Am J Surg 1996;171:27–31. Weaver ML, Ashton JG, Zemel R. Treatment of colorectal liver metastases by cryotherapy. Semin Surg Oncol 1998;14:163–170. Zhou XD, Tang ZY. Cryotherapy for primary liver cancer. Semin Surg Oncol 1998;14:171–174. Seifert JK, Finlay I, Armstrong N, Morris DL. Thrombocytopenia and cytokine release following hepatic cryosurgery [abstract]. Aust N Z J Surg 1998;68:526. Goldblum SE, Hennig B, Jay M, et al. Tumor necrosis factor alpha–induced pulmonary vascular endothelial injury. Infect Immun 1989;57:1218–1226. Ferrari-Baliviera E, Mealy K, Smith RJ, Wilmore DW. Tumor necrosis factor induces adult respiratory distress syndrome in rats. Arch Surg 1989;124:1400–1405.
Radiofrequency ablation (RFA) is gaining increasing approval in the treatment of unresectable primary and metastatic hepatic tumors.1-4 Hepatic resection remains the standard of care for resectable lesions.5-7 Ongoing international experience with the use of RFA for unresectable tumors has clearly demonstrated its safety over cryotherapy. Initial encouraging reports of the efficacy of RFA have prompted many users to adopt this procedure over
cryoablation for these unresectable hepatic tumors. With introduction of progressively larger RFA arrays and capability to generate larger ablations, combined with the applicability of minimally invasive surgery to RFA, the relative advantages previously held by cryotherapy appear to be waning. In addition, cases of major systemic inflammatory complications, including hepatic and renal failure, coagulopathy, and syndromes similar to adult respiratory distress syndrome (ARDS), are observed in patients treated with cryotherapy, particularly in patients who undergo ablation of large volumes of liver (⬎30% of total hepatic volume).8-11 Appearance of “cryoshock”-like clinical entities12 has not been reported in patients treated with RFA, and this absence of marked inflammatory response has remained
No competing interests declared.
Received January 18, 2002; Revised April 22, 2002; Accepted June 25, 2002. From the Departments of Surgery and Molecular Genetics and Microbiology, University of Florida College of Medicine, Gainesville, FL; and Surgical Service, Malcolm Randall Veterans Affairs Medical Center, Gainesville, FL. Correspondence address: Scott R Schell, MD, PhD, FACS, Department of Surgery, University of Florida College of Medicine, 1600 SW Archer Rd, Box 100286, Gainesville FL 32610-0286.
© 2002 by the American College of Surgeons Published by Elsevier Science Inc.
774
ISSN 1072-7515/02/$21.00 PII S1072-7515(02)01333-9
Schell et al
Vol. 195, No. 6, December 2002
Cytokine Release after Radiofrequency Ablation
775
Table 1. Patient Demographics Patient Age no. (y) Gender
Diagnosis
1 2 3 4 5
50 67 66 85 72
F M M M M
CRC CRC HCC with marked cirrhosis CRC HCC with marked cirrhosis
6 7 8 9 10 11 12 13 14 15 16 17
40 73 73 63 51 58 63 69 48 77 73 63
F M M M F M M F F F M M
Carcinoid CC CRC GI stromal tumor Leiomyosarcoma CRC HCC CRC BC CRC CC CRC
Length of stay (d)
Liver volume ablated*
2 3 2 7 30
14.7% 36.8% 35.8% 30.7% 19.8%
3 3 4 6 5 3 4 3 5 5 3 2
37.9% 37.3% 9.7% 18.8% 55.3% 35.8% 44.9% 31.5% 46.5% 48.6% 58.3% 48.1%
Complications
None None None Ileus, urinary tract infection Upper GI hemorrhage, bacteremia, hepatic failure, death on postoperative day 30 None None None None, underwent transfusion for low preoperative hematocrit None None None None None None None, underwent transfusion for low preoperative hematocrit None
*As percentage of calculated liver mass. BC, breast carcinoma; CC, cholangiosarcoma; CRC, colorectal carcinoma metastases; GI, gastrointestinal; HCC, hepatocellular carcinoma.
intriguing from the perspective of how RFA affects innate immunity and production of pro- and antiinflammatory cytokines after treatment. Previous reports have examined elaboration of proinflammatory cytokines after a variety of surgical procedures, ranging from minimally invasive surgery13-16 to major vascular surgical reconstruction.17 In these studies, elevation of proinflammatory cytokines in the patients’ plasma was directly associated with major systemic responses. No previously published reports have evaluated production of pro- or antiinflammatory cytokines in patients undergoing RFA treatment. Several authors have published reports using rat models of RFA and cryotherapy and have examined cytokine production, lung inflammation, and the histopathology of lung injury after treatment.18-21 These studies demonstrated a direct association between cryotherapyinduced lung injury and NF-B activation and production of NF-B–dependent cytokines, tumor necrosis factor (TNF)-␣, and macrophage inflammatory protein-2. The purpose of this study was to examine production of pro- and antiinflammatory cytokines in patients undergoing RFA treatment for unresectable primary and metastatic hepatic tumors.
METHODS Patient recruitment
Patients undergoing RFA were recruited from Shands Hospital at the University of Florida, Gainesville, FL. The Institutional Review Board at our institution approved experimental protocols, and signed informed consent was obtained from all patients before enrollment. The enrollment and study of research patients at our institution are in strict compliance with the Helsinki Declaration of 1976. Seventeen patients undergoing RFA for unresectable primary or metastatic liver tumors agreed to blood sampling before and for 48 hours after RFA treatment for measurement of pro- and antiinflammatory cytokines. Patient demographics
Seventeen patients (11 men, 6 women) with unresectable primary (hepatocellular carcinoma, three patients) or metastatic (colorectal carcinoma, eight patients; cholangiocarcinoma, two patients; sarcoma, two patients; breast, one patient; carcinoid, one patient) hepatic tumors were treated in this study. Patient age ranged from 40 to 85 years (mean⫾SEM, 64.2⫾2.8 years). Patient demographics, diagnoses, and postoperative length of stay are described in Table 1.
776
Schell et al
Cytokine Release after Radiofrequency Ablation
Study protocol
Preoperative imaging using either CT or MRI was obtained for all patients before they underwent laparotomy and abdominal exploration. After the presence of extrahepatic metastases was excluded, the size, location, geometry, and histology of each hepatic lesion were confirmed using intraoperative ultrasound and ultrasoundguided core biopsy. After biopsy confirmation of histology, each metastasis was treated using LeVeen-type monopolar RFA arrays and either the RF2000 or RF3000 RFA generator (Radiotherapeutics, Inc, Palo Alto, CA), using techniques described previously.1,4 Each lesion was treated with at least two cycles of RFA, pursuant to our standard ablation protocol, with the first RFA cycle continued until either maximal tissue impedance was obtained, or until a total of 15 treatment minutes had elapsed. After a 2-minute “cool-down” period, a second cycle of RFA was performed using the time and impedance criteria just described. Hepatic tumors that were larger than our probe size were divided into overlapping quadrants, and each quadrant was treated separately using the protocol just described. A margin of normal liver of at least 1 cm was included with each ablation, and ablated margins were verified using intraoperative ultrasonography. The estimated volume of liver treated using RFA was calculated as a percentage of total hepatic volume using methods described previously, with liver mass (in grams) calculated as (patient weight in kilograms)/70)⫻(4/ 7)⫻1,000.22 Every treated lesion was measured using intraoperative ultrasound, and volume of ablated tissue was calculated after adding a 1-cm margin of normal liver to the largest radius of each lesion. The volumes of all ablated lesions were then summed and divided into the total liver volume and expressed as a percentage of total liver volume. The weight-based methodology used for liver volume calculations has been shown to be highly accurate in autopsy and subsequent clinical series, but this methodology might tend to somewhat overestimate actual liver volumes ablated. Postoperatively, patients were admitted to the surgical oncology service and were closely monitored for fever and signs or symptoms of sepsis. Patients were encouraged to walk on the first postoperative day and were discharged home when they were able to tolerate a regular diet. After discharge, hepatic CT imaging was performed 1 week and 1 month after RFA to confirm success of treat-
J Am Coll Surg
ment and to compare effective ablated hepatic volume with that calculated during the RFA procedures. Ablated liver volume as assessed using postoperative CT imaging was observed to be highly concordant with the volumes calculated at the time of operation. Blood collection for cytokine assays and ex vivo stimulation
Blood samples (7 mL) were obtained after induction of anesthesia and immediately before commencement of RFA to provide internal baseline controls for each patient. Additional blood samples (7 mL) were obtained after completion of the final RFA treatment and thereafter at timed intervals (2 minutes and 1, 2, 4, 8, 24, and 48 hours after RFA). All blood samples were collected with dipotassium ethylenediaminetetraacetic acid, were immediately placed in ice, and were promptly centrifuged at 1,300 g at 4°C for 15 minutes; the plasma fraction was separated, aliquoted, and stored at ⫺70°C until analysis, as described previously by our laboratories.17 Additional blood samples (7 mL) were obtained for examination of ex vivo stimulated cytokine production after induction of anesthesia—to provide baseline levels for each patient—and at 24 hours and 48 hours after the final RFA treatment. Samples were collected and immediately placed in ice, and the whole blood used for ex vivo cytokine production was measured after stimulation with lipopolysaccharide, using the protocol described previously.23,24 Biochemical analyses
Freshly thawed plasma samples were assayed by ELISA for TNF␣, interleukin [IL]-1, IL-1ra, IL-6, IL-8, IL10, and the TNF shed receptors p55 and p75, as described previously by our laboratories.17 The sensitivity of the TNF-␣, IL-6, IL-8, IL-10, p55, and p75 assays varied from 32 to 128 pg/mL, as has been reported previously.17,25,26 Values obtained below the detection limits of an individual assay were treated as 0 for purposes of statistical analysis. Whole-blood samples were stimulated with LPS overnight at 37°C at final concentrations of 1 ng/mL or 1 g/mL using methods described previously.23 Whole blood was assayed by ELISA for TNF␣, IL-1, IL-1ra, IL-6, IL-8, IL-10, and TNF shed receptors p55 and p75 at 0, 24, and 48 hours after LPS stimulation using the identical assay procedures described earlier.
Vol. 195, No. 6, December 2002
Schell et al
Statistical analyses
All data are expressed as the mean ⫾ SEM. Baseline measurement data were obtained for each patient after anesthesia induction and before commencement of RFA. Within-patient changes from baseline in plasma cytokine levels were analyzed using one-way ANOVA to compare the means at all the time points, with any overall significant differences subjected to Tukey’s honestly significant difference test. Differences were considered significant if the two-sided p value was less than 0.05. RESULTS Postoperative hospital length of stay ranged from 2 to 30 days (mean 5.3⫾1.6 days). No patients developed systemic inflammatory syndromes equivalent to either cryoshock or ARDS. One patient (patient 5) had a history of marked cirrhosis and preoperative variceal bleeding. After RFA treatment, this patient experienced recurrence of variceal bleeding on the second postoperative day, which progressed to irreversible hepatic failure refractory to transjugular portosystemic shunting and systemic bacteremia. The patient died on the 30th postoperative day. Of the remaining 16 patients, none exhibited postoperative hypotension. Postoperative ileus and urinary tract infection developed in one patient (patient 4), who was discharged on the seventh postoperative day without further complications. There were no other complications. Pyrexia (ⱖ38.5°C) developed in three patients (patients 2, 5, and 8) during the first 48 hours after completion of RFA, but these temperature elevations were not associated with notable leukocytosis or hemodynamic perturbation; they resolved spontaneously with ambulation and incentive spirometry and did not require treatment with antibiotics. Postoperative hepatic CT imaging at 1 week and 1 month after RFA confirmed successful RFA treatment in all patients. When compared with baseline levels measured in each patient at induction of anesthesia, RFA treatment did not induce pronounced changes in plasma levels of the cytokines IL-1, IL-8, IL-10, or TNF-␣ (Fig. 1A). After RFA, plasma IL-1ra, IL-6, p55, and p75 concentrations were only modestly stimulated (Fig. 1B), but these levels did not reach statistical significance. There were no associations observed between the elevations in plasma cytokine appearance and the development of postoperative pyrexia occurring in patients 2, 5, and 8. RFA treatment did not alter the production of pro-
Cytokine Release after Radiofrequency Ablation
777
and antiinflammatory cytokines as measured by ex vivo LPS stimulation of whole-blood samples before and after RFA treatment. Figure 2A shows stimulated IL-1, IL-8, IL-10, and TNF-␣ production before RFA treatment and at 24 and 48 hours after treatment. Figure 2B shows LPS-stimulated IL-1ra, IL-6, p55, and p75 during the same time period. Although a modest decrease in TNF-␣ production was noted 24 to 48 hours after RFA (4.4 ng/mL versus 13.4 ng/mL control), the differences in concentration did not reach statistical significance, and no enhancement or inhibition of the other cytokines in response to LPS stimulation was observed. DISCUSSION These preliminary studies suggest that RFA does not induce a significant plasma appearance of the proinflammatory cytokines IL-1, IL-1ra, IL-6, IL-8, IL-10, TNF-␣, or the TNF shed receptors p55 and p75 in human patients undergoing RFA for unresectable primary or metastatic liver tumors. It appears that RFA does not activate production of the same proinflammatory cytokines elaborated after cryotherapy of similar liver tumors19 and does not appear to mimic the same degree of inflammatory response observed after the minimally invasive surgical procedure laparoscopic cholecystectomy.13,16 In addition, RFA does not substantially alter cytokine production after LPS stimulation of whole blood obtained from these patients. Although patients in this study were treated with RFA to achieve ablation volumes equivalent to those obtained in patients developing cryoshock after cryotherapy, our observations about these patients’ cytokine production are concordant with our institution’s observations and experience1,3,4 of patients undergoing large-volume RFA who do not exhibit symptoms consistent with the cryoshock phenomenon after ablation. With a combined total of nearly 65,000 new cases of colorectal hepatic metastases and hepatocellular carcinomas each year, and with only 25% to 35% of patients presenting with resectable lesions, there remains a considerable need for additional treatment modalities for these patients.7,27 In addition, although prospective randomized trials comparing hepatic resection and RFA for resectable lesions are still forthcoming, the observed benefits of hepatic resection—with 5-year survivals of 30% to 35% after complete resection of three to five lesions28—raise the question whether hepatic resection
778
Schell et al
Cytokine Release after Radiofrequency Ablation
J Am Coll Surg
Figure 1. Cytokine concentrations after radiofrequency ablation (RFA) treatment. (A) Interleukin (IL)-1, IL-8, IL-10, and tumor necrosis factor (TNF)-␣. (B) IL-6, IL-1ra, p55, and p75. Cytokine concentrations were assessed at time of anesthesia induction (A.I.), after RFA probe placement (RFA Start), on completion of all RFA treatments (RFA Stop), and at the time points indicated thereafter. Each data point represents the mean for the series. Error bars are ⫾ 1 SEM.
provides adequate outcomes for the risks and resources consumed. Past reports of cryotherapy, although demonstrating efficacy and versatility in treating a variety of tumor sizes, types, geometries, and intrahepatic locations,9,10,29-33 have shown that cryotherapy is also complicated by risks of coagulopathy, renal and respiratory failure, and other major
morbidities at rates, in some series, equivalent to those described for hepatic resection.8,11,12 There is increasing evidence that hepatic Kupffer cells are important participants in inflammatory responses after hepatic injury or hypoperfusion. Seifert and colleagues34 have recently demonstrated that cryoablation results in TNF-␣ and IL-6 release in a rat model. Previ-
Vol. 195, No. 6, December 2002
Schell et al
Cytokine Release after Radiofrequency Ablation
779
Figure 2. Cytokine production by LPS-stimulated whole blood after radiofrequency ablation (RFA) treatment. (A) Interleukin (IL)-1, IL-8, IL-10, and tumor necrosis factor (TNF)-␣. (B) IL-6, IL-1ra, p55, and p75. Patient whole blood was collected before RFA (0 hours) and 24 and 48 hours after completion of RFA treatment. Collected samples were treated with low (1 ng/mL) or high (1 g/mL) concentrations of lipopolysaccharide, and cytokine production was assessed. Each data point represents the mean for the series. Error bars are ⫾1 SEM.
ous studies have shown that direct endothelial damage and clinical syndromes equivalent to ARDS arise after IV infusion of TNF-␣, identifying this cytokine as the likely culprit in the previously characterized cryoshock phenomenon.12,35,36 The absence of proinflammatory cytokine production after RFA treatment does not appear related to the
body’s inability to elaborate these cytokines, because cytokine production by the patient’s endotoxin-stimulated whole blood was not observed to be notably decreased. Instead, we propose that the most likely explanation for our observations is based in the mechanism of tumor destruction achieved by using RFA, specifically, that the thermal injury created by RFA results in coagulation,
780
Schell et al
Cytokine Release after Radiofrequency Ablation
cellular protein denaturation, and subsequent cell death. Accordingly, it is reasonable to expect that adjacent Kupffer cells and their released cytokines would be subjected to the same thermal coagulation and so be rendered inactive. In contrast, cryotherapy involves freezing using liquid nitrogen–cooled probes, and cellular lysis and necrosis result from intracellular ice crystals forming as a result of the freeze-thaw cycles. Cellular necrosis, unlike apoptosis, is generally associated with an inflammatory response, particularly in organs rich in tissue macrophages. In addition, because few cytokines are inactivated by freezing, cytokines elaborated during cryoablation would remain active to induce local and systemic effects. Together, these preliminary findings suggest that hepatic tumor ablation after RFA-induced thermal coagulation has much lower systemic inflammatory potential than cryotherapy while yielding equivalent degrees of tissue ablation. The advent of RFA for the treatment of unresectable hepatic metastases has provided an additional therapeutic alternative for patients with unresectable hepatic tumors while avoiding the risks associated with cryoablation and hepatic resection. Author Contributions
Study conception and design: Schell, Wessels, Abouhamze, Moldawer, Copeland Acquisition of data: Schell, Wessels, Abouhamze, Moldawer, Copeland Analysis and interpretation of data: Schell, Wessels, Abouhamze, Moldawer, Copeland Drafting of manuscript: Schell, Wessels, Abouhamze, Moldawer, Copeland Critical revision: Schell, Wessels, Abouhamze, Moldawer, Copeland Statistical expertise: Schell, Wessels, Abouhamze, Moldawer, Copeland Supervision: Schell, Moldawer REFERENCES 1. Wessels FJ, Schell SR. Radiofrequency ablation treatment of refractory carcinoid hepatic metastases. J Surg Res 2001;95:8– 12. 2. Rose DM, Allegra DP, Bostick PJ, et al. Radiofrequency ablation: a novel primary and adjunctive ablative technique for hepatic malignancies. Am Surg 1999;65:1009–1014. 3. Siperstein AE, Rogers SJ, Hansen PD, Gitomirsky A. Laparoscopic thermal ablation of hepatic neuroendocrine tumor metastases. Surgery 1997;122:1154–1155.
J Am Coll Surg
4. Curley SA, Izzo F, Delrio P, et al. Radiofrequency ablation of unresectable primary and metastatic hepatic malignancies: results in 123 patients [see comments]. Ann Surg 1999;230:1–8. 5. Fong Y, Kemeny N, Paty P, et al. Treatment of colorectal cancer: hepatic metastasis. Semin Surg Oncol 1996;12:219–252. 6. Kavolius J, Fong Y, Blumgart LH. Surgical resection of metastatic liver tumors. Surg Oncol Clin North Am 1996;5:337– 352. 7. Fong Y, Blumgart LH. Surgical options in the treatment of hepatic metastases for colorectal cancer. Curr Probl Surg 1996; 35:336–413. 8. Sarantou T, Bilchik A, Ramming KP. Complications of hepatic cryosurgery. Semin Surg Oncol 1998;14:156–162. 9. Stewart GJ, Preketes A, Horton M, et al. Hepatic cryotherapy: double-freeze cycles achieve greater hepatocellular injury in man. Cryobiology 1995;32:215–219. 10. Haddad FF, Chapman WC, Wright JK, et al. Clinical experience with cryosurgery for advanced hepatobiliary tumors. J Surg Res 1998;75:103–108. 11. Bagia JS, Perera DS, Morris DL. Renal impairment in hepatic cryotherapy. Cryobiology 1998;36:263–267. 12. Seifert JK, Morris DL. World survey on the complications of hepatic and prostate cryotherapy. World J Surg 1999;23:113– 114. 13. Bellon JM, Manzano L, Larrad A, et al. Endocrine and immune response to injury after open and laparoscopic cholecystectomy. Int Surg 1998;83:24–27. 14. Berguer R, Bravo N, Bowyer M, Ferrick D. Measurement of intracellular gamma-interferon, interleukin-4, and interleukin-10 levels in patients following laparoscopic cholecystectomy. J Invest Surg 2000;13:161–167. 15. Helmy SA, Wahby MA, El-Nawaway M. The effect of anaesthesia and surgery on plasma cytokine production. Anaesthesia 1999;54:733–738. 16. Lemaire LC, van der Poll T, van Lanschot JJ, et al. Minimally invasive surgery induces endotoxin-tolerance in the absence of detectable endotoxemia. J Clin Immunol 1998;18:414–420. 17. Welborn MB, Oldenburg HS, Hess PJ, et al. The relationship between visceral ischemia, proinflammatory cytokines, and organ injury in patients undergoing thoracoabdominal aortic aneurysm repair. Crit Care Med 2000;28:3191–3197. 18. Washington K, Debelak JP, Gobbell C, et al. Hepatic cryoablation-induced acute lung injury. Histopathologic findings. J Surg Res 2001;95:1–7. 19. Chapman WC, Debelak JP, Wright Pinson C, et al. Hepatic cryoablation, but not radiofrequency ablation, results in lung inflammation. Ann Surg 2000;231:752–761. 20. Chapman WC, Debelak JP, Blackwell TS, et al. Hepatic cryoablation-induced acute lung injury: pulmonary hemodynamic and permeability effects in a sheep model. Arch Surg 2000;135:672–673. 21. Blackwell TS, Debelak JP, Venkatakrishnan A, et al. Acute lung injury after hepatic cryoablation: correlation with NF-kappa B activation and cytokine production. Surgery 1999;126:518– 526. 22. DeLand FH, North WA. Relationship between liver size and body size. Radiology 1968;91:1195–1198. 23. van der Poll T, Coyle SM, Moldawer LL, Lowry SF. Changes in endotoxin-induced cytokine production by whole blood after in vivo exposure of normal humans to endotoxin. J Infect Dis 1996;174:1356–1360.
Vol. 195, No. 6, December 2002
Schell et al
24. Fong YM, Marano MA, Moldawer LL, et al. The acute splanchnic and peripheral tissue metabolic response to endotoxin in humans. J Clin Invest 1990;85:1896–1904. 25. Welborn MB III, Douglas WG, Abouhamze Z, et al. Visceral ischemia-reperfusion injury promotes tumor necrosis factor (TNF) and interleukin-1 (IL-1) dependent organ injury in the mouse. Shock 1996;6:171–176. 26. Van Zee KJ, Kohno T, Fischer E, et al. Tumor necrosis factor soluble receptors circulate during experimental and clinical inflammation and can protect against excessive tumor necrosis factor alpha in vitro and in vivo. Proc Natl Acad Sci U S A 1992;89:4845–4849. 27. American Cancer Society. Cancer facts and figures. Atlanta, GA: American Cancer Society; 1998. 28. Hughes KS, Simon R, Songhorabodi S, et al. Resection of the liver for colorectal carcinoma metastases: a multiinstitutional study of patterns of recurrence. Surgery 1986; 100:278–284. 29. Adam R, Majno P, Castaing D, et al. Treatment of irresectable
30. 31. 32. 33. 34. 35. 36.
Cytokine Release after Radiofrequency Ablation
781
liver tumours by percutaneous cryosurgery. Br J Surg 1998;85: 1493–1494. Seifert JK, Morris DL. Prognostic factors after cryotherapy for hepatic metastases from colorectal cancer. Ann Surg 1998;228: 201–208. Shafir M, Shapiro R, Sung M, et al. Cryoablation of unresectable malignant liver tumors. Am J Surg 1996;171:27–31. Weaver ML, Ashton JG, Zemel R. Treatment of colorectal liver metastases by cryotherapy. Semin Surg Oncol 1998;14:163–170. Zhou XD, Tang ZY. Cryotherapy for primary liver cancer. Semin Surg Oncol 1998;14:171–174. Seifert JK, Finlay I, Armstrong N, Morris DL. Thrombocytopenia and cytokine release following hepatic cryosurgery [abstract]. Aust N Z J Surg 1998;68:526. Goldblum SE, Hennig B, Jay M, et al. Tumor necrosis factor alpha–induced pulmonary vascular endothelial injury. Infect Immun 1989;57:1218–1226. Ferrari-Baliviera E, Mealy K, Smith RJ, Wilmore DW. Tumor necrosis factor induces adult respiratory distress syndrome in rats. Arch Surg 1989;124:1400–1405.