Hyperthermic Isolated Limb Perfusion with TNF α and Cisplatin in the Treatment of Osteosarcoma of the Extremities: A Feasibility Study in Healthy Dogs

Purpose. The feasibility of hyperthermic isolated limb perfusion (HILP) with tumor necrosis factor-α (TNFα ) and cisplatin for the management of osteosarcoma was studied in the canine model. Methods. During seven perfusions in six healthy mongrel dogs (weight 32±2 kg) technical aspects of HILP under mild hyperthermia (39– 40℃) were studied. In five experiments HILP was performed with TNFα alone (0.5 mg/l extremity volume), and in two experiments TNFα was combined with cisplatin (25 mg/l extremity volume). During the perfusions physiological parameters were monitored and TNFα and total cisplatin concentrations were determined. Results. Perfusion conditions (pH, PCO2 , PO2, flow and pressure) remained within physiological ranges.Three dogs died within 24 h despite a sublethal systemic concentration of TNFα that leaked from the perfusion circuit. Three dogs were terminated; one dog after the second experiment in accordance with Dutch ethical rules; one dog showed an invagination of the small bowel resulting in an ileus; one dog because of necrosis of the perfused limb. Conclusions. This feasibility study in healthy dogs demonstrated that HILP with TNFα and cisplatin was associated with a high mortality rate and does not allow us to treat dogs with spontaneous osteosarcoma with TNFα and cisplatin HILP. Therefore, an alternative model should be used in the search for the ideal combination of perfusion agents for limb sparing treatment in human osteosarcoma.


Introduction
Osteosarcom a is the m ost frequent prim ary m alignant bone tum or in hum ans. Until the 1970s the m ost com m on approach to the m anagem ent of localized osteosarcom a was surgical resection, am putation or radiation therapy. 1 D uring the last decades a de® nite role for neoadjuvan t high-dose m ethotrexate and cisplatin-based polychem otherapy was established.
1± 4 T he p otential lo cal tum or effect of system ically adm inistered cisplatin, however, is lim ited due to the nephrotoxicity and ototoxicity of cisplatin. T herefore an attem pt was m ade to increase the local effect of cisplatin without increasing system ic toxicity by using hyperthermic isolated regional lim b perfusion (HILP) with cisplatin in dogs w ith spontaneous osteosarcom a. 5 T hese studies showed an acceptable locoregional toxicity, im proved functional outcom e at 6 and 12 weeks, and a steadily im proving radiological picture. However, the histological results were modest, with none of the dogs showing a com plete response at 6 weeks after perfusion. T he sam e experience was found in patients with sarcom as of soft tissue and bone treated w ith cisplatin H ILP. 6 Results of recent publications and of our own experience w ith a new perfusion m odality, which com bines tum or necrosis factor-a (TN F a ) and m elphalan in patients w ith recurrent m elanom a or soft tissue sarcom a, are very prom ising. 7,8 H owever, in six of eight evaluable patients with unresectable osteosarcom a of the lower lim b treated w ith T N F a an d m elp halan H IL P, histological evaluation revealed m oderate results with 80% necrosis in three patients, 50± 60% necrosis in two patients and < 50% necrosis in one patient. After TN F a and m elphalan H ILP, lim b spar ing surgery was possible in six patients. 9 As cisplatin is one of the m ost active chem otherapeutics in the treatm ent of osteosarcom a, it seems worthwhile to investigate the results of HILP with T N Fa and cisplatin. W ith the high frequency of occurrence in dogs, canine osteosarcom a is a useful m odel for evaluation of new treatment regim ens in humans as rapid case accrual and rap id tim e to reach m easu rable en d p oin ts are possible. 10 T he canine osteosarcom a therefore appears to be a valid m odel for studying the potential treatment of H IL P with T N Fa and cisplatin in the local treat m en t o f osteo sarco m a o f th e extrem ity in hum ans. To establish optim al HILP conditions using TN Fa and cisplatin for local tum or control in dogs bearing osteosarcom a, a feasibility study in healthy dogs was undertaken.

Dogs
During seven experim ents in six healthy m ongrel dogs with a m ean average weight of 32 ± 2 kg and a m ean age of 6 ± 1 years, different asp ects of HILP with TN Fa and cisplatin were studied. Preoperatively, all dogs w ere tho rou ghly clin ically evaluated at the Central Anim al Facility of the University of G roningen.T he study was approved by the Anim al Welfare Comm ittee of the Faculty of M edicine of the G roningen U niversity.

Anaesthetics
The dogs fastened for 12 h and were anaesthetized with thiopental (30 m g/kg body w t., i.v.) (Pentothal, Abbott, Am stelveen, T he N etherlands) and, after m u scle rela xatio n w ith p a n cu ro n iu m b ro m id e (0.08 m g/kg body w t., i.v.) (Pavulon, Organon, O ss, The N etherlands), the dogs were ventilated (Ohm eda Modulus 2) with a mixture of O 2 and iso¯urane. T he oxygen concentration in the gas m ixture was continuously m easured by m ean s of an oxygen analyzer (Ohm eda M odulus 2) and m inute volum es (4± 6 l/ min) were adjusted to maintain an end-expirator y CO 2 concentration of 4± 5% (Siem ens C O 2 analyzer 930). The dogs were placed in the supine position on a heated m attress to m aintain their norm al body tem perature of 38Ê C. 11 D uring the operations all dogs were given about 2 l of glucose 5% via a cephalic or internal jugular vein. C entral arterial pressure was recorded as well as an EC G and diuresis.

Operation and perfusion techniques
D uring anaesthesia the volum e of the extrem ity was measured using Archim edes' rule (1.7± 2 l). T he iliac vessels were exposed under sterile conditions and collateral vessels were clipped. C annulas were inserted into the artery (Bardic, 14± 18 F ) and vein (Bardic, 14± 18 F ). B o th can nu las were con n ected to an extracorporeal circuit consisting of an occlusive roller pump, a cardiotomy reservoir and a bubble oxygenator with heat-exchanger. A tourniquet m ade of nylon was placed around the base of the extrem ity, using a pin in the bone and a bandage around the m iddle to com plete the isolation of the limb from the system ic circulation. T he perfusate consisted of 350 m l 5% dextran 40 in glucose 5% (Isodex, Pharm acia AB, U ppsala, Sweden), 250 ml red blood cells ( determ ined in the regional and system ic circulation at 0, 5, 15, 30, 45, 60, 75 and 90 m in by EL ISA and am eless ato m ic ab so r p tio n sp ecto p ho to m etr y (FAAS), respectively. The perfusion time was 1 h, followed by wash-o ut of the extrem ity with 3 l of Isodex. Tourniquet, cannulas and clips were then rem oved and the incisions in the vessels repaired. P ro tam in e hyd ro ch lo r id e (H o ff m a n L a R o ch e, M ijdrecht, The N etherlands) was adm inistered, to neutralize heparin, in a ratio of 1:1 to the initial dose of heparin. All dogs were closely observed for at least 24 h. N o anti-in¯am m atory or analgesic drugs were adm inistered during follow-up.
All dogs were followed for local and system ic side effects of TN Fa and cisplatin perfusion, as well as sur vival. Table 1 show s the characteristics of the seven experim ents in six dogs. D uring the experim ents conditions for perfusions (pH , PCO 2 , PO 2 ) were kept within the physiolo gical ranges, as in hum an perfusions. Figure  1 shows the¯ow, blood pressure, perfusion pressures, weight gain or loss of the extracorporeal circuit and temperature during 60 m in of perfusion in the seven experim ents. In the ® rst ® ve experim ents, only T N Fa was adm inistered to the perfusion circuit. In the last two experim ents cisplatin was added. Figure   2 illustrates the T N Fa concentrations (m ean ± SEM ) in the perfused lim b as well as in the system ic circulation of the dog during perfusion and afterwards. Peak T N F a concentration s in the perfu sed lim b were 650 ± 158 ng/m l, and in the system ic circulation of the dog they were 37 ± 15 ng/m l. The peak system ic concentrations in the dog were in the sam e range as those of TN Fa and m elphalan HILP used in the treatment of hum ans at our institute. 16 Figure 3 shows the m easured total cisplatin values in the last two experim ents. D uring the experim ents we were not able to perform any leakage m onitoring by m eans of radionuclear detection techniques which are used in the clinical perfusion setting. T herefore leakage was calculated afterwards according to Stehlin with the am ount of blood in the dogs estimated at 69 m l/kg b o d y w eigh t. 17 C a lcu lated lea ka ge valu es are sum m arized in Table 1. T hree dogs died within 24 h: the ® rst two during the T N Fa experim ent; the third after T N Fa and cisplatin perfusion. Postm ortem exam ination of these an im als d id n o t p rov id e an y m acro sco pic o r m icroscopic evidence to explain the cause of death. T hree dogs were term inated; two due to treatm ent com plications. O ne dog showed an invagination of the sm all bowel, resulting in an ileus and another was terminated 1 week after TN Fa and cisplatin perfusion because of necrosis of the perfused lim b. T he third dog was term inated after the second experim ent in accordance with D utch ethical rules.

Discussion
In the treatment of osteogenic sarcom a a distinction can be m ade between system ic therapy and locoregion al treatm en t. H igh-d o se m eth otre xate-b ased system ic chemotherapy is prim arily adm inistered in order to eradicate possib le m icrom etastatic disease, and its use was a m ajor breakthrough in the clinical treatm ent of osteosarcom as in the 1970s.
1,2 Today, about 60% of patients with resectable prim ary tumors and no m etastases at the time of the initial diagnosis will be cured. 1 T he prim ary objective in locoregional treatment is to prevent local recurrence and allow lim b salvage procedures in an attempt to preserve lim b function. N ew surgical techniques and the developm ent of en do prosthetic m ater ials, cou pled w ith system ic neoadjuvant chem otherapy, have offered less radical surgery for 40± 80% of patients with osteosarcom a since the 1980s.
1,18 Procedures that increase tum or necrosis of the prim ary tum or, w ith reduction of viable tumor cells and tumor volum e, could contribute to lim b preservation strategies. Since its ® rst use, cisplatin has been one of the m ost effective chemotherapeutic agents and has been incorporated in m ost system ic treatment regim ens for osteosarcom a. A recen t attem pt to overco m e its n ephrotoxic an d ototoxic lim itations by administering cisplatin in HILP in the treatment of spontaneous canine osteosarcom a was histologically m odest. 5 Prom ising results of recent publications and our own experience with a new com bination perfusion m odality (TN Fa and m elphalan) for recurrent m elanom a or soft tissue sarcom a, but m oderate histological results in patients with osteosarcom a, prom pted us to investigate the com bination of TN Fa and cisplatin in H ILP for osteosarcom a. 7± 9 Since endothelial cells are supposed to play a key role in the working m echanism of T N F, osteosarcom as w ith a high extent of tum or vessels are of particular interest.
Before application of TN F a and cisplatin HILP in humans and client-owned osteosarcom a-bearing dogs, the present feasibility study was perform ed in norm al healthy dogs. D espite sufficient experience in H ILP in dogs as well as in hum ans, an unexpected high m ortality rate was encountered. Although there was no m ortality related to the operation, three dogs died w ith in 24 h a fter p erfu sio n (50 % ). T h is d irect postoperative m ortality could not be explained by a surplus of system ical leakage of TN F. In the experim ent, the dog w ith the highest leakage and, as a  26 ; Grade I, no reaction, objectively and subjectively; Grade II, slight erythema, oedem a or loss of sensation; G rade III, considerable erythema or oedema with some blistering, slight functional disturbances; G rade IV, extreme epiderm iolysis and/or obvious damage to the deep tissues causing de® nite functional disturbances; Grade V, reaction that might necessitate amputation.
consequence, the highest system ic T N Fa concentrations, survived im m ediately postoperatively, and the dog with the lowest leakage (lowest system ic T N Fa concentrations) died within 24 h after perfusion. N o correlation between leakage and m ortality rate could be established. M axim al leakage encountered in these experim ents was 33% , this corresponds to 330 m g T N Fa given system ically per dog; since the average dog weighs 33 kg, the dose of T N Fa that reaches the system ic circulation of the dog is sublethal (10 m g/ kg). 14 Although only sublethal doses of T N Fa leaked to the system ical circu lation , the clin ical picture resem bled respon se s o bse r ved w ith leth al do se s ( > 1 00 m g /k g) , ch a ra cter iz ed b y p ro g re ssive selectively prom oted the in vitro perm eability of the blood± brain barrier to C D D P without disrupting the tight junctions. 24 An improved penetration of cisplatin into the interstitial space due to a higher permeability of the vascular wall, com bined with an increased cellular cisplatin accum u lation and D N A addu ct form ation, could explain the observed necrosis of the lim b in this in vivo m odel with the cisplatin dose used, w hich was previously non-toxic. T he observed m or tality and m orbidity that we encountered in this canine study was sim ilar to the experience of W ithrow and colleagues (unpublish ed obser vations). T he present results in norm al elderly m ongrel dogs indicate that treatm ent of dogs with spontaneous osteosarcom a using TN Fa and cisplatin HILP is not appropriate. Future research could focus on postoperative m onitoring and care in dogs after TN F a H ILP; perhaps a better alternative for testing the effect of T N Fa with cisplatin HILP is the use of the rat osteosarcom a m odel described by M anusam a et al., 25 since rats are m uch less susceptible to TN F a than dogs.

Acknowledgm ents
The advice and technical assistance, in the G ron-