Cross section of the Infusaid (Infusaid Corporation, Norwood, Mass) pump.
Cross section of the pump by Arrow International Inc (Reading, Pa). A, Correct needle placed for refilling pump. B, Special needle placement for bolus injections.
A, Pump pocket being created in the right side of a midline incision. B, Pocket fully created to be the size of the pump.
Overview of the pump catheter placement in the gastroduodenal artery. A, Catheter shown inside the gastroduodenal artery with knots laid proximal and distal to the fist stay. B, The second stay is tied with the suture around the distal gastroduodenal artery.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Kemeny MM. The Surgical Aspects of the Totally Implantable Hepatic Artery Infusion Pump. Arch Surg. 2001;136(3):348–352. doi:10.1001/archsurg.136.3.348
The design of a totally implantable hepatic artery infusion pump in 1969 made the use of hepatic artery infusion feasible and practical as a treatment for patients with hepatic neoplasms. The implantable pumps could function for long periods and reliably infuse a measured quantity of drug into the hepatic artery in a continuous fashion. This enabled oncologists to give much higher doses of chemotherapy directly into the blood supply of the tumors as well as to use a continuous infusion schedule.
In the 1970s many studies examined the response of hepatic metastases from colorectal primary tumors to hepatic artery infusion (HAI) using the implantable pump. Early unrandomized studies reported high response rates and even complete responses in patients with hepatic metastases from colorectal primary tumors.1,2 However, during the early part of the decade many patients who were treated with HAI (not study patients) suffered numerous complications. Several centers started prospective randomized studies of HAI vs control treatment (usually systemic chemotherapy) to see if there was a real benefit to this treatment over the systemic therapies available for patients with hepatic metastases from colorectal cancers.3-7 All of these studies reported significant positive responses to HAI, but since many of the studies had crossover designs whereby patients who failed systemic therapy received HAI, it was difficult to be sure there was a survival benefit of HAI alone. After 10 years of testing it took a meta-analysis of all of the studies to show that there indeed was a survival benefit for patients with metastases from colorectal primary tumors who were treated with HAI.8 Furthermore, the decade of study delineated which of these patients would benefit the most from treatment. Patients with metastatic colorectal cancer who had liver disease only, had less than 70% of their liver involved with metastases, and had a good performance status responded best to this modality.
Complications of pump placement and pump therapy were also better understood. For instance, it became clear that the gallbladder had to be removed at the time of pump placement. Currently, pump placement requires an abdominal operation. The morbidity and mortality of this procedure should be close to zero. New combinations of chemotherapy and appropriate selection of patients should provide at least a 70% response rate with this form of therapy for the treatment of hepatic metastases from colorectal primary tumors, which is at least a 2-fold better response rate than that of current systemic chemotherapies.9,10
The basic design of the original pump was a 2-chambered unit made of titanium. One chamber contained the drug and could be accessed from outside of the pump. The other chamber, called the charging fluid chamber, was filled with Freon and totally sealed off. The pump was supplied with mechanical energy that was continuously refilled. Fluid entered the drug chamber by means of a percutaneously placed needle. The filling of the chamber would push out the bellows that compressed the charging chamber. The compressed Freon would then exert its energy by pushing up on the diaphragm of the device and in turn slowly push out fluid through the catheter of the pump (Figure 1). The activity of the Freon at body temperature was well regulated and would be constant every day until the pump went dry. The first pumps used were made by the Infusaid Corporation (Norwood, Mass) and had a 50-mL reservoir. They also had a side port by which the surgeon could inject directly into the catheter.
In recent years the pump has been redesigned by different manufacturers with variances in the placement of the side port. The pump made by Arrow International Inc (Reading, Pa), for example, features the port as part of the pump mechanism and it can only be accessed from the central needle inlet by a special catheter needle (Figure 2).
All patients must undergo a hepatic arteriogram, including an arteriogram of the celiac axis and the superior mesenteric artery, prior to placement of the pump. In the normal anatomy, the common hepatic artery is a branch of the celiac axis. The first branch off the common hepatic artery should be the gastroduodenal artery followed by the bifurcation of the hepatic artery into right and left hepatic arteries. The superior mesenteric arteriogram is needed to be sure that there are no accessory right hepatic arteries branching off the superior mesenteric artery.
A study of arteriograms of 100 consecutive patients who needed pump placement showed that only 50% of patients had normal hepatic arteriovasculature.11 The most common deviation was that of a right hepatic artery branching off the superior mesenteric artery. This was either an accessory to the right hepatic artery, which bifurcates off the common hepatic artery, or a replaced right hepatic artery (if there was no right hepatic artery off the common hepatic artery). An accessory or replaced left hepatic artery was the next most frequent anomaly. This could be seen during the celiac angiogram as a branch off the left gastric artery. A final arterial anomaly, and one that caused difficulty in placing the pump, was a trifurcation, whereby the gastroduodenal artery and the bifurcation of the right and left hepatic arteries branched off at exactly the same place. This was troublesome for pump placement since the flow from the catheter into the gastroduodenal artery could go entirely into 1 branch instead of equally into both branches.
When the arterial anatomy is normal, the catheter is placed into the gastroduodenal artery. If there is an accessory or replaced artery, usually the accessory artery is tied off and the catheter is still placed into the gastroduodenal artery. If both arteries are replaced and there is no gastroduodenal artery, then a catheter is placed directly into the biggest one of the replaced vessels and the other is tied off.
Placement of the surgical incision depends on the operation the patient is having. If the operation involves only the placement of the HAI pump, then a midline incision should be used. If the procedure for placing the pump is in addition to a hepatic resection, then a bilateral subcostal incision, or chevron incision, should be used. The procedure for pump placement includes a cholecystectomy if the gallbladder is still present. Open cholecystectomy can be done with any operative technique and is performed to remove the gallbladder, which can be a source of postoperative problems during infusion of chemotherapy. When the pumps were first being used, at least 30% of patients who did not have their gallbladders removed experienced acalculous cholecystitis during HAI. Once this was recognized it became standard to remove the gallbladder at the time of pump placement.
The hepatic artery should be localized in the area of the gastrohepatic ligament. Often, there are lymph nodes that lie on top of the hepatic artery in this area. The more superficial lymph nodes should be dissected, removed, and sent to a pathology laboratory for a frozen section. If they are positive for tumor, the pump will probably not be beneficial for the patient because this signifies extrahepatic disease. Once the hepatic artery can be seen, it should be dissected to find the gastroduodenal branch (the first branch located caudally). The gastroduodenal branch must also be cleanly dissected so that all of the surrounding small lymphatic branches are obliterated. The common hepatic artery should be cleaned off distally from the gastroduodenal takeoff to the bifurcation of the right and left hepatic arteries. The artery must be dissected circumferentially to be sure that there are no small branches in the posterior region going to the pancreas.
After the gastroduodenal artery is cleanly dissected, it should be ligated as distally as possible. A hemostat attached to the distal tie can aid in visualizing the posterior area of the artery, by tenting up on the ligated structure. Once the whole area is dissected, the common hepatic artery should be freely mobile to such an extent that the operating surgeon can easily put his or her finger behind the bifurcation of the hepatic and gastroduodenal arteries. The common hepatic artery should be cleanly dissected at least 2 cm distally and proximally to the gastroduodenal bifurcation.
Once these maneuvers are completed, the pocket for the pump should be made. The pump will be placed superficial to the fascia, generally in the left upper quadrant of the abdomen. This area is optimal for the pump because it does not get in the way of subsequent computed tomography or magnetic resonance imaging scans of the liver used to evaluate whether the liver metastases are responding to pump therapy. The pocket is made by grabbing the fascia with Kocher clamps, stretching it, and then using a cautery to create a pocket directly above the fascia from the incision laterally. The opening of the pocket should be made slightly smaller than the pump so the pump will be secure, but large enough so that the pump can be slipped into it completely, with no area of the pump being visible. Once the pocket has been created, cautery should be used to ensure that all bleeders are cauterized and that the pocket is completely dry.
The pump should be kept in a warmer throughout the first part of the operation and not brought to the field until the pocket is completed. If the pump is not warm (ie, at least body temperature), there will be no pressure in the catheter and consequently there may be retrograde flow from the artery into the pump catheter. This may cause clotting of the catheter. When the pocket is ready for implantation the pump should be primed with a heparin solution of 30 mL of 1000 U/mL heparin (30 000 U of heparin). This solution should be kept in the warmer too before it is placed into the pump. The pump needs to be drained of all fluid and then filled with the 30-mL heparin solution. To access the drug chamber, a noncoring needle must be used (Huber needle), generally of a 19- or 20-gauge diameter. Finally, the pump catheter should be flushed using a heparin solution of 100 U/mL. The pump is now ready to be put into the pocket.
A hole is made in the fascia of the pocket so that the catheter can be pulled through it. It is easiest to do this by poking a right-angled clamp through the fascia from inside the abdomen to the pocket, and placing it in the center of the pocket. The catheter of the pump is then passed into the jaws of the clamp, pulled through the fascia, and dropped inside the abdomen. The body of the pump should then be placed in the pocket, but not sewn in, so that during the catheter insertion the pump is within the walls of the pocket obtaining warmth from the body. It is important not to let the pump cool down too much below body temperature. Next, the catheter is inserted into the gastroduodenal artery.
The pump catheter is prepared by cutting the catheter at an angle about 1 cm distal to the first stay ring. For the Arrow pump it is important to see the heparin solution beading at the catheter opening. If this is not seen, the pump has to be put in a warmer at the operating table to assure that it is functioning. Once the beading is seen, the operation can continue.
The gastroduodenal artery is prepared by placing a vascular clamp on it near the takeoff point of the gastroduodenal and hepatic arteries. With countertraction on the suture ligature, the gastroduodenal artery is tented up using fine forceps, and an arteriotomy is made with a pair of fine Pott scissors. Using a plastic catheter introducer, the catheter is slipped through the arteriotomy into the artery. The vascular clamp is opened enough to let the catheter pass through the clamp's jaws and then closed again, holding the catheter in place. The catheter is then tied in place using a 2-0 silk ligature around the main part of the gastroduodenal artery with the catheter inside it and numerous knots are made on this suture until the knotted section reaches the first stay. At that point, the 2-0 silk is tied on the distal side of the first stay so that the stay is situated between the ties on the artery and the ties on the stay (Figure 3A). Then the 2-0 silk that was used to tie off the gastroduodenal artery is passed around the second stay of the catheter and ties the catheter a second time (Figure 3B).
To check for correct placement of the catheter, 5 to 10 mL of 50% fluorescein is injected into the side port of the pump, followed with 10 mL of a 100-U/mL heparin solution. All of the lights in the operating room should be turned off before the injection of the fluorescein and an ultraviolet lamp (Wood lamp) should be used to visualize the liver. Both lobes of the liver must light up. The stomach should not light up. Once this is done, the pump is put back into the pocket (Figure 4).
If the whole liver does not light up, an unrecognized accessory artery is probably present and must be found and ligated. If the defect is in the left lobe, the artery can be found in the gastrohepatic ligament between the esophagus and the hepatic artery. If the defect is in the right lobe, the artery will be found behind the portal vein in the posterior aspect of the porta hepatis.
The pump is secured in the pocket by sewing the fascia and one of the stay loops together using a 0 chromic suture. The pump pocket is closed with interrupted 0 chromic sutures between the fascia and the subcutaneous tissue. This closure should be watertight.
The closure of the abdominal wound should be done with particular care, because if there is a wound infection, it could spread to the pump pocket and the pump would have to be removed.
Patients can be fed immediately. No nasogastric suction is necessary. Antibiotics should be continued for a few days since an infection might necessitate removal of the pump. To check pump placement postoperatively, a nuclear medicine scan is taken using macroaggregate albumin labeled with technetium Tc 99m injected into the side port. The liver should light up as a result.
Patients can start receiving pump treatments as soon as the medical oncologist wishes, usually commencing 1 week after the implantation day. The pump should be refilled every 2 weeks. If for some reason the patient is not ready for therapy in 2 weeks, the pump should be filled with the heparin solution again.
The pump can flip inside the pocket, making it impossible to access. This is very rare but more common with Arrow pumps because they do not have a side port keeping it in place. If this happens early in the course it can be flipped mechanically without entering the pocket. Later on the pocket might have to be entered surgically to flip the pump over and sew it in place. This is an extremely rare occurrence but should be considered if there are access problems.
The major problems with HAI are not surgical. They include gastritis, duodenitis, and biliary sclerosis. Biliary sclerosis may require an endoscopically placed biliary stent.
There have been cases of the catheter eroding through the wall of the duodenum when the pump has been in place for more than a year. These patients usually complain of abdominal pain leading to an endoscopy. Surgery is required to remove the catheter and repair the duodenum.
If the catheter is placed too low in the gastroduodenal artery, the chemotherapy from the pump will pool in the gastroduodenal artery instead of going directly into the hepatic artery. The high concentration of chemotherapy combined with the pooling can cause the wall of the artery to erode. This not only can cause bleeding but chemotherapy can extravasate into the abdomen and cause pain. This is usually diagnosed by a computed tomography scan that shows a hematoma in the area of the gastroduodenal artery. The pump must be emptied and no further fluid should be placed in it. This will allow the catheter to clot and may also stop the bleeding. The patient should be monitored to be sure there is no further bleeding. The pump can be removed at any time after this event. If the bleeding continues, an abdominal exploration may be necessary to remove the catheter and tie off the gastroduodenal artery.
If the hepatic artery is not dissected cleanly during the placement of the pump, some of the branches to the pancreas may remain and pancreatitis might follow. At least 1 case of a pancreatic pseudocyst has been reported. In this case the pseudocyst was filled with contrast from the pump catheter.12 For problems of this sort, HAI must be stopped permanently.
Pump pocket infections occur rarely during the course of treatment. At the first sign of infection (erythema over the pump pocket), systemic antibiotics need to be started. If the infection does not resolve, the pump needs to be moved to a new location in a newly created pocket. The old pocket should be opened and drained.
The treatment of hepatic metastases from colorectal cancer has occupied the attention of surgical oncologists more than any other metastatic disease. Combinations of ablative surgery, chemotherapy, and HAI have expanded the possibilities of curative treatment for this usually fatal disease. The use of HAI became especially prominent with the introduction of the totally implantable pump. However, the proper placement of the HAI pump is not as simple as placing a port into an artery. There are many peculiarities about the pump itself and the hepatic artery anatomy that make familiarization with the procedure necessary. The follow-up and complications of HAI need to be recognized and dealt with quickly and accurately, for late recognition of a complication such as biliary sclerosis can lead to fatal consequences.
Corresponding author: M. Margaret Kemeny, MD, Division of Surgical Oncology, HSC T-18 Room 60, SUNY Stony Brook, Stony Brook, NY 11794-8191 (e-mail: email@example.com).
Create a personal account or sign in to: