Technical complications of continuous intra-arterial chemotherapy with 5-fluorodeoxyuridine and 5-fluorouracil for colorectal liver metastases☆☆☆

Patient characteristics 

All 180 patients (74 females, 106 males) had histologically proven carcinoma of the colon or rectum metastatic to the liver (Table I). Of these 180 patients, 88 had synchronous and 92 had metachronous metastasis. Twenty-seven patients had extrahepatic disease as well. These patients were accepted for intra-arterial therapy contrary to the eligibility criteria, because the extent of extrahepatic metastasis was small and localized.

Table I. Patient characteristics

	No. of patients	n = 180
	Female/male	n = 74/106
	Synchronous metastases	n = 88 (49%)
	Metachronous metastases	n = 92 (51%)
	Number of metasases
	≤4 metastases	n = 31 (18%)
	>4 metastases	n = 149 (82%)
	Pattern of hepatic involvement
	Localized (1 lobe)	n = 30 (17%)
	Disseminated	n = 150 (83%)
	Infusion devices
	Ports	n = 110 (61%)
	Pumps	n = 70 (39%)

Thirty-one patients (18%) had 4 or less hepatic metastases, the remaining 149 (82%) patients had more than 4 hepatic metastases before intra-arterial chemotherapy. In the majority of patients (n = 150) these metastases affected the entire liver, whereas in 30 patients only 1 lobe was involved (Table I).

Of the 180 patients included in this study, 110 patients (61%) received an intra-arterial port and 70 patients (39%) an intra-arterial infusion pump.

Eligibility criteria 

Before regional chemotherapy, each patient provided written informed consent. To be eligible for intra-arterial chemotherapy, patients had to have unresectable liver metastases. Furthermore, the liver had to be the only site of metastasis. Serum bilirubin levels had to be less than 3 mg/dL, and the Karnofsky performance score had to be above 60%. If more than 80% of the liver had been replaced by tumor or a colorectal recurrence was present, patients were excluded from this study. Liver cirrhosis and thrombosis of the portal vein were additional exclusion criteria, as were ascites and the presence of a second primary tumor.

Catheter and infusion devices 

Commonly available implantable silicone catheters were used with a cuff at the tip to facilitate consistent placement. They were connected either to an implanted drug pump (model 400, Infusaid Inc, Providence, RI) or to a port (Celsite, Braun Co, Melsungen, Germany). The implanted pump had a 50-mL reservoir enabling a constant flow of 2 to 4 mL/day. This facilitated a continuous infusion over 30 days. A separate side port at the pump was available for bolus injections.

Because all treatment regimens required continuous infusion, ports were connected to an extracorporeal, portable drug pump using standardized port-a-cath puncture sets for safe and easy puncture.

Preoperative evaluation 

Preoperative evaluation included a complete history, thorough physical examination, and routine laboratory tests (complete blood count, partial thromboplastin time, prothrombin time) and biochemical profiles (bilirubin, creatinine, alkaline phosphatase, hepatic transaminases). All patients had a chest radiograph and a computed tomography scan of the abdomen and pelvis, and most of the patients had an additional ultrasound examination of the abdomen. Recurrence of the colorectal primary tumor was excluded by colonoscopy in all patients. Preoperative angiography was performed to illustrate the vascular anatomy and to demonstrate the feasibility of catheter implantation.

Operative technique 

Laparotomy was performed by a right subcostal incision. If not previously performed, all patients underwent cholecystectomy to avoid chemical cholecystitis. Thereafter, the implantation of the arterial catheter was performed by the Watkins' technique11: After careful exposition of the common hepatic artery and its branches, the gastroduodenal artery was identified and ligated 1.5 cm behind its runoff from the hepatic artery. After dissection of the gastroduodenal artery proximal to this ligation, the catheter was inserted and carefully advanced to the junction of the common hepatic and gastroduodenal arteries without extending into the lumen. The catheter was fixed using bidirectional ligation on either side of its cuff with nonresorbable sutures. Further branches of the hepatic artery distal to the origin of the gastroduodenal artery (ie, right gastric artery) and up to 2 cm proximal to the gastroduodenal artery had to be ligated to avoid chemoperfusion of the stomach or the small bowel. To demonstrate the exact position of the catheter and to exclude misperfusion, fluorescein was injected into the catheter. The correct perfusion of the liver was proven by the Wood's lamp. After the exact position had been confirmed, the catheter was connected to the heparin-filled infusion device. The port systems were subcutaneously placed onto the right costal arch. The infusion pumps were implanted into a subcutaneous pocket in the right lower abdominal wall.

Despite common variations in the anatomy of the hepatic artery, this standard technique can be used in nearly all cases. Further manipulations such as transpositions of the hepatic arteries or the insertion of 2 catheters or pumps should not be performed. If an additional hepatic artery is present, the nondominant artery can be ligated, because intrahepatic shunts rapidly reconstitute the inflow after ligation of a variant hepatic artery and provide sufficient blood supply to the liver within 5 days.12 If a catheter implantation is not possible in these ways, the patient is ineligible for this treatment and should not receive intra-arterial chemotherapy by an implanted port or pump.

Postoperative care 

To ensure early detection in the event of misperfusion, postoperative perfusion controls of the implanted catheter were conducted regularly. Perfusion scintigraphy with 99Technetium-labeled human albumin was conducted by the port or side port of the infusion pump approximately 1 week after catheter implantation to confirm the exact position of the catheter before the first cycle of chemotherapy. Perfusion controls were performed in 3-month intervals by scintigraphy or angiography during regional chemotherapy.

Treatment regimens 

The first cycle of chemotherapy was started 2 weeks after the implantation of the infusion devices. Patients were treated according to 6 different treatment regimens with continuous infusion of 5-FU or FUDR (Table II).

Table II. Treatment protocols used for continuous regional chemotherapy of colorectal liver metastases

		Protocol	Period of application (days)	Duration of cycles (days)	Duration of treatment pauses (days)	No. of patients
	Protocol 1	FUDR 0.2 mg/kg BW/day (i.a.)	14	28	14	56
	Protocol 2	FUDR 0.2 mg/kg BW/day (i.a.)	14	28
		+			14	29
		5-FU 30 mg/kg BW/day (i.v.)	3	28
	Protocol 3	FUDR 1-1.2 mg/kg BW/day (i.a.)	5	28	23	40
	Protocol 4	FUDR 1.2-1.7 mg/kg BW/day(i.a.)
		+	5	21	16	19
		Leucovorin 30 mg/kg BW/day (i.a.)
	Protocol 5	5-FU 200 mg/m2/day (i.a.) +
		Leucovorin 200 mg/kg BW/day (i.a.)	14	28	14	12
		+
		FUDR 0.05 mg/kg BW/da (i.a.)
	Protocol 6	5-FU 350-1000 mg/m2/day (i.a)
		+	5	28	23	24
		Leucovorin 20-500 mg/m2/day (i.a.)

BW, Body weight; i.a., intra-arterial.

On the first day of the treatment cycle, the subcutaneously implanted infusion pumps were filled with 50 mL saline solution containing the appropriate dosage of the chemotherapeutic agent. During treatment pauses, the infusion pumps were filled with a heparin solution to avoid thrombotic occlusion of the intra-arterial catheter.

Patients with subcutaneous port systems received continuous intra-arterial chemotherapy by an extracorporeal infusion pump filled with the chemotherapeutic agent and connected to the port, on the first day of the treatment cycle. After removing these infusion pumps on the last day of the treatment cycle, the port was irrigated with a heparin solution.

Statistics 

Complication-free survival was defined as time from catheter implantation until occurrence of a complication or death and calculated according to the Kaplan-Meier method.13 Proof of significance was calculated by log-rank test.

Technical complications 

Technical complications for implanted pumps and ports were analyzed separately. We differentiated between those complications affecting the intra-arterial catheter (catheter-associated complications) and those affecting the infusion device (pump or port). Catheter-associated complications included thrombotic occlusion of the catheter or the hepatic artery, dislocation, and leakage of the catheter. In addition, we defined therapy-relevant complications resulting in cessation of treatment.

The overall complication rate was 62% (68 of 110) for patients with port systems versus 41% (29 of 70) for patients with implanted pumps (Table III).

Table III. Complications during continuous intra-aterial chemotherapy

		Port	Pump
	Reversible catheter occlusion	9 (8%)	-
	Irreversible catheter occlusion*	19 (17%)	3 (4%)
	Dislocation of the catheter tip*	15 (14%)	7 (10%)
	Leakage of the catheter*	9 (8%)	-
	Thrombosis of hepatic artery*	5 (5%)	2 (3%)
	Pump failure*	-	5 (7%)
	Local infection*	2 (2%)	4 (6%)
	Rupture of puncture membrane*	2 (2%)	-
	Puncture not possible	4 (4%)	-
	Hematoma	2 (2%)	-
	Seroma	1 (1%)	-
	Skin necrosis	-	3 (4%)
	Skin perforation	-	2 (3%)
	Insufficient fixation	-	2 (3%)
	Filling with less volume possible	-	1 (1%)
	Total	68/110 (62%)	29/70 (41%)
	*Therapy-relevant complications.

Catheter-associated complications 

Catheter-associated complications occurred in 57 patients (52%) after port implantation. There were 28 episodes of thrombotic catheter occlusion, the most frequently observed complication. In addition, thrombosis of the hepatic artery occurred in 5 port systems. Of these 33 thrombotic complications, 24 (73%) were observed in treatment protocols 3 and 4, using high-dose FUDR. The use of low-dose FUDR or 5-FU (protocols 1, 2, 5, 6) resulted in 9 (27%) thrombotic episodes. Nine of all thrombotic occlusions were reversible by local lysis or high-pressure heparin injections.

Catheter-associated complications were observed in 12 patients (17%) after pump implantation. All catheter-associated complications in patients with pumps were therapy-relevant, because they were all primarily irreversible (Table III).

Device-associated complications 

Device-associated complications occurred in 11 patients (10%) with ports: the puncture of the port was not technically achievable in 4 patients (4%), a hematoma occurred in 2 patients (2%), a rupture of the port membrane and an infection of the port occurred in 2 patients (2%) each, and a seroma was observed in 1 patient (1%) after the surgical procedure.

Complications affecting the implanted pump occurred in 17 patients (24%). There was no hematoma or seroma, but skin perforation was documented in 2 patients (3%), and skin necrosis in 3 patients (4%). As a result of an insufficient fixation of the pump, puncture and filling of the pump was more difficult in 2 patients (3%).

Therapy-relevant complications 

Complications that could not be treated, requiring the cessation of treatment or a change in infusion device, were categorized as therapy-relevant complications. They were observed in 52 patients (47%) after port implantation and in 21 patients (30%) after pump implantation.

An irreversible thrombotic occlusion occurred in 19 patients (17%) after port implantation. In addition, the high-pressure injection of heparin for irrigating the occluded catheters frequently caused catheter rupture, resulting in leakage in 9 port systems (8%) and dislocation of the catheter tip in 15 episodes (14%), resulting in misperfusion, pain, or extravasation. Rupture of the port membrane and an infection of the port occurred in 2 episodes (2%), respectively.

The most frequently observed therapy-relevant complication (n = 7) after pump implantation was dislocation of the catheter tip. Malfunction of the implanted pump occurred in 5 episodes (7%), and a pump pocket infection was observed in 4 episodes (6%). Clotting of the catheter occurred in 3 patients (4%), and thrombotic occlusion of the hepatic artery in 2 patients (3%). All of these complications were primarily irreversible.

Complication-free survival 

The median complication-free survival was significantly longer (P = .0016) for patients with pumps (12.2 months) than for those with ports (7.3 months) (Fig 1).

View Large Image Download to PowerPoint

Fig. 1. Complication-free survival after implantation of infusion devices. Broken line, intra-arterial port (n = 110 patients); solid line, intra-arterial pump (n = 70 patients).