Objectives: The aims of this study were to determine if angiographic findings can be used to predict successful nonoperative therapy of splenic injury and to determine if coil embolization of the proximal splenic artery provides effective hemostasis.
Methods: Splenic injuries detected by diagnostic imaging between 1981 and 1993 at a level I trauma center were prospectively collected and retrospectively reviewed after management by protocol that used diagnostic peritoneal lavage, computed tomography (CT), angiography, transcatheter embolization, and laparotomy. Computed tomography was performed initially or after positive diagnostic peritoneal lavage. Angiography was performed urgently in stabilized patients with CT-diagnosed splenic injuries. Patients without angiographic extravasation were treated by bed rest alone; those with angiographic extravasation underwent coil embolization of the proximal splenic artery followed by bed rest.
Results: Patients (172) with blunt splenic injury are the subject of this study. Twenty-two patients were initially managed operatively because of associated injuries or disease (11 patients) or because the surgeon was unwilling to attempt nonoperative therapy (11 patients) and underwent splenectomy (17 patients) or splenorrhaphy (5 patients). One hundred fifty of 172 consecutive patients (87%) with CT-diagnosed splenic injury were stable enough to be considered for nonoperative management. Eighty-seven of the 90 patients managed by bed rest alone, and 56 of 60 patients treated by splenic artery occlusion and bed rest had a successful outcome. Overall splenic salvage was 88%. It was 97% among those managed nonoperatively, including 61 grade III and grade IV splenic injuries. Sixty percent of patients received no blood transfusions. Three of 150 patients treated nonoperatively underwent delaye!
d splenectomy for infarction (one patient) or splenic infection (two patients).
Conclusions: (1) Hemodynamically stable patients with splenic injuries of all grades and no other indications for laparotomy can often be managed nonoperatively, especially when the injury is further characterized by arteriography. (2) The absence of contrast extravasation on splenic arteriography seems to be a reliable predictor of successful nonoperative management. We suggest its use to triage CT-diagnosed splenic injuries to bed rest or intervention. (3) Coil embolization of the proximal splenic artery is an effective method of hemostasis in stabilized patients with splenic injury. It expands the number of patients who can be managed nonoperatively.
Diagnostic peritoneal tap and lavage [diagnostic peritoneal lavage (DPL)] is a most reliable method of identifying hemoperitoneum associated with intraperitoneal injuries.  Although most surgeons would agree that some of these injuries do not require treatment and, indeed, that some are better off left alone, DPL is not particularly helpful in determining which patients may be treated nonoperatively. A positive DPL does not determine the source or activity of bleeding, often requiring surgeons to perform exploratory celiotomy for definitive diagnosis and therapy.
The concept of nonoperative management of the injured spleen has been advanced considerably with the development of computed tomography (CT). Abdominal CT is also a reliable method for identifying hemoperitoneum and staging splenic injury, and an effective test for excluding associated injuries that might require laparotomy. 
Computed tomography, however, also has drawbacks that limit the number of patients and the types of splenic injury that can be treated nonoperatively. It does not often determine whether bleeding is still persistent or has stopped. The CT appearance of splenic injury does not predict with great certainty the success of nonoperative therapy.
The authors have applied an approach to the diagnosis and treatment of splenic injury that incorporates both DPL and CT for diagnosis, uses arteriography to triage identified splenic injury, and attempts transcatheter arterial embolization of the proximal splenic artery as the initial method of definitive hemostasis in patients who are thought to be at risk for persistent or delayed hemorrhage from their splenic injury. We review the results of our methods of nonoperative treatment of splenic injury.
PATIENTS AND METHODS
The authors prospectively collected and reviewed all patients who had splenic injury identified by CT between the years of 1981 and 1993. Demographics, diagnostic evaluations, extent of abdominal injury, mode of therapy, and outcome were examined.
In 1979, two of the authors (S.J.A.S., G.W.S.) began a study to assess the ability of arteriography to predict which stable patients with hemoperitoneum detected by DPL could be managed without celiotomy and by bed rest alone. Panangiography, including abdominal aortography (and, when appropriate, abdominal or pelvic aortography), and celiac and mesenteric arteriography were performed before exploratory laparotomy in stable patients with hemoperitoneum detected by DPL.
The study was terminated before its completion when abdominal CT was reported to be a reliable comprehensive alternative to DPL. It was clear to us that angiography would be inferior to CT as a screening test. We chose CT as an alternative to DPL in stabilized patients with suspicion of abdominal trauma.
Several things, however, were learned from this aborted study: (1) angiography was successfully completed without complication or death in all patients; (2) minor hepatic injuries, resulting in hemoperitoneum but requiring no therapy at laparotomy, were not detected by arteriography; (3) when splenic arteriography showed arterial contrast extravasation, splenic injuries were bleeding at laparotomy; (4) splenic injuries that did not extravasate during arteriography did not bleed at laparotomy until the spleen was manipulated; and (5) mesenteric arterial spasm and extravasation correlated with operative findings of mesenteric injury.
At our institution, CT replaced DPL in 1982. Despite the recognized value of CT, however, it became clear within 24 months that it was technically impossible for the institution to replace DPL with CT in all patients with suspected abdominal trauma. The location of the CT scanner three floors above and one building away from the emergency department and the great demand placed on our single scanner by other practitioners were obstacles that we could not overcome. It was decided that DPL would be reinstituted as the screening triage tool for most patients with suspected abdominal trauma.
Nonetheless, the value of CT in allowing the nonoperative management of selected hemodynamically stable patients with abdominal visceral injury was recognized, and we felt that to return to mandatory exploration of patients with a positive DPL would be a disservice to some.
Instead of performing mandatorily exploratory celiotomy for patients with a positive DPL, CT, with some exceptions (Table 1), was then performed to evaluate further hemoperitoneum detected by DPL in patients who remained stabilized. This methodology has been shown to be safe and effective. 
Table 1. Reasons CT was not performed after positive DPL.
Method of CT
The CT scans were performed on second-generation or third-generation scanners, depending on the available technology at the time. The scanners were located within the main radiology department where technologists and radiology residents were present 24 hours a day. Patients were transported from the emergency department to the radiology department by a nurse or a resident surgeon. Most patients who had a positive DPL were monitored by pulse oximetry and had a radial artery line placed for continuous arterial pressure recording before transfer to the imaging center.
All patients underwent gastrointestinal opacification before CT scanning by the oral or nasogastric administration of 1000 to 2000 mL of a 3% iodinated contrast medium solution begun approximately 30 to 60 minutes before initiating scanning. Vascular and visceral enhancement were obtained by the intravenous administration of an ionic 60% iodinated contrast medium solution immediately before scanning. Contiguous 8- to 10-mm-thick sections were imaged from the higher diaphragm to the lowest portion of the ischium.
Action after Normal CT
Patients who did not have DPL performed and whose CT showed no hemoperitoneum or visceral injury were discharged from the emergency department or after one night of hospital observation, unless concurrent injury required further hospitalization. Patients who had a positive DPL and a normal CT were more enigmatic; these patients were observed or underwent celiac and mesenteric arteriography, depending on clinical appearance and the responsible surgeon.
Action after CT Showed Splenic Trauma
In general, patients with splenic injury identified by CT underwent urgent angiography, including celiac and superior mesenteric arteriography if they remained hemodynamically stabilized, and CT did not show associated injury that required laparotomy.
Angiography and Embolization for Splenic Trauma
Cut film analog angiography without digital imaging was performed in most patients. A few of the more recent patients have had digital subtraction angiography performed as well. Right femoral arterial access was used primarily. Most studies included serial films in multiple projections after transcatheter arterial injection of ionic iodinated contrast medium.
If no arterial extravasation was visible on celiac or splenic arteriography, patients were admitted to the intensive care unit for initial observation without additional intervention. Nasogastric suction, invasive monitoring, serial hemoglobin measurements, and physical examination were performed.
If arterial extravasation was seen on celiac arteriography, selective splenic artery catheterization was attempted. The catheter was positioned approximately 2 cm beyond the origin of the dorsal pancreatic artery and proximal to the first pancreatica magna artery. Steel spring coils of slightly larger diameter than the splenic artery were injected to occlude the splenic artery at that point.
Specific and focal splenic arterial branch embolization with small pledgets of gelfoam was performed only in those uncommon patients whose arteriogram showed that the extravasated contrast medium had extended outside the capsule into the peritoneum. In such situations, 0.1-mL increments of gelfoam pledgets were directed by flow into the periphery of the splenic vasculature until the extravasation was no longer seen. Embolizations of the liver, kidney, or hypogastric vessels were performed when arteriographic extravasation was seen from these areas. Antibiotics were not administered routinely after embolization. The patient was then treated by bed rest and observation in a fashion similar to those patients who did not have arterial extravasation.
In this plan of care, celiotomy was considered for failure of nonoperative management. The predefined parameters for exploration are listed in Table 2. An algorithm diagram is supplied in Figure 1.
Table 2. Indications for exploratory laparotomy after CT or angiography.
Figure 1. Algorithm for the management of blunt abdominal trauma. SICU, surgical intensive care unit.
One hundred seventy-two patients were reviewed. The patients' histories and presentations are summarized in Table 3; the degree of abdominal injury is summarized in Table 4. The CT grading system for splenic injury was that advocated by the American Association for the Surgery of Trauma. The outcome of therapy is summarized in Table 5, and complications are listed in Table 6.
Table 3. Patient history and presentation.
Table 4. Analysis of abdominal injuries.
Table 5. Outcome of therapy.
Table 6. Complications.
The Student-Newman-Keuls test was used to analyze the statistical variance of a number of parameters between those treated by bed rest, by splenic artery occlusion followed by bed rest, or by primary laparotomy. There was no significant difference (p > 0.050) in age, admitting systolic, diastolic, and mean blood pressure or pulse rate. Injury Severity Score was slightly higher in the embolized and explored groups than in those treated by bed rest alone (p > 0.050), but this was not significant (p > 0.050). The incidence of bloody paracentesis was higher in the group undergoing primary laparotomy. We are unclear whether this increase in laparotomy represents a higher degree of injury or was a preconceived bias based on the surgical practice encountered during our early experience. Although cell counts were higher in patients undergoing embolization or laparotomy, there was no statistically significant differences (p > 0.050) in lavage !
The CT appearance of splenic injury was most severe in the patients who underwent embolization and lowest in those treated by bed rest alone. The CT appearance of more severe injuries correlated with but did not predict the angiographic findings.
The presence of extravasation on arteriography was uncommon in grade I injuries (12%), and most patients with grade I injuries (82%) did not require embolization or operation. One-third of grade II injuries and one-half of grade III injuries showed extravasation and underwent embolization. On the other hand, 29% of grade IV injuries showed no extravasation.
The absence of extravasation on arteriography was a reliable indicator of successful nonoperative therapy, and none of these patients had persistent or recurrent hemorrhage or required delayed laparotomy. Three patients, treated by bed rest based on CT findings without undergoing arteriography, had recurrent or persistent hemorrhage. They did not undergo arteriography immediately after CT because their injuries were thought to be trivial and to not require further diagnostic evaluation. Laparotomy in one patient with falling hematocrit revealed a nonbleeding grade I splenic injury that was managed by splenorrhaphy and a nonbleeding laceration of a cirrhotic liver. Another patient in group I underwent delayed arteriography after recurrent hemoperitoneum, and this showed arterial extravasation for which splenic artery occlusion was performed. This patient did not bleed further and was successfully managed nonoperatively. The third patie!
nt had arteriography that showed extravasation, but the radiologist chose not to perform embolization of the splenic artery at that time. When hemorrhage recurred, coil embolization was performed, and the patient had an uneventful outcome.
The overall percentage of patients treated nonoperatively was 88%. The rate of splenic salvage was 97% among those initially treated nonoperatively.
Among those treated initially by operation after CT diagnosed a splenic injury, 11 of 22 patients were explored because of the preference of the attending surgeon. These patients occurred early in our experience when some of the surgeons felt uncomfortable with the concept of nonoperative therapy for splenic injury. Most of these patients did not undergo splenic arteriography. The remaining 11 patients underwent laparotomy because of pre-existing disease, severe intracranial injury, or associated intra-abdominal injuries.
There were three deaths related to splenic injury. One elderly patient with a grade IV splenic injury and a pelvic fracture with sacroiliac dislocation had severe concurrent coronary artery disease and previous laparotomy. He sustained cardiac arrest during CT. Another patient, previously mentioned, with severe pre-existing hepatic failure and ascites had a grade I splenic injury and a liver laceration. Splenorrhaphy was successful, but the patient died of hepatic failure. The third patient died after splenic artery coil occlusion. This Jehovah's Witness refused DPL, transfusion of red blood cells or clotting factors, and laparotomy.
The experience with these last two patients suggests to us that nonoperative therapy may be less likely to be successful if the clotting mechanism is deficient, because splenic artery occlusion does not directly stop bleeding, but rather facilitates the body's own reparative processes. We do not have proof of this opinion.
Coil occlusion was relatively safe. Embolization was technically successful in each patient, although there were some technical difficulties, once resulting in a complication. Three patients had initially unsatisfactory placement of coils, but two of these misplacements were improved or rectified by other techniques of interventional radiology without untoward effects on patient outcome.
One patient had errant embolization of the coil out of the splenic artery into the aorta, and it traveled distally and lodged in the popliteal artery. This coil was removed by snare technique, and the embolization was then completed with another coil, without complication. In another patient, the tail of the coil protruded into the celiac axis. This was also removed by snare technique and replaced with a smaller coil placed more distally in the middle portion of the splenic artery, also without sequelae. In a third patient, a coil of insufficient diameter was placed in the proximal splenic artery, but because it was smaller than the midsplenic artery, it embolized distally into the hilum of the spleen. Snare retrieval was unsuccessful, and the spleen infarcted because the coil lodged beyond the collateral circuits of the splenic artery. Splenectomy was performed and Staphylococcus epidermis was cultured from the organ. Angiographic di!
ssection occurred in two patients, but this was asymptomatic and nonocclusive, and was not considered a complication. Angiography was repeated in one of these patients, and this showed healing of the splenic artery dissection without aneurysm formation or stenosis.
Splenic artery coil occlusion was augmented by embolization of pledgets of surgical gelatin in two patients. Both became hypotensive during the procedure before coil embolization, and their arteriogram showed active extrasplenic extravasation. Pledgets of gelfoam of 2 to 3 mm diameter were flow-directed from the proximal splenic artery into the periphery of the spleen before the splenic artery was occluded with coils. The hemodynamic response was dramatic and very rapid after particulate embolization. Both patients responded to transfusion without recurrent shock. Neither developed clinical signs of sepsis or splenic infarction.
Complications were uncommon. Two patients (one previously mentioned) who underwent coil occlusion developed sepsis within the first week. Both underwent exploration and splenectomy. Perisplenic hematomas were infected with Staphylococcus epidermis and Staphylococcus aureus. Both patients did well after splenectomy. A third patient with respiratory failure underwent laparotomy and splenectomy 17 days after trauma. Scattered infarcts in a viable spleen were seen, but cultures were negative. Two other patients had CT-guided aspiration of a liquefied hematoma because of fevers. Culture was negative in both, and both recovered without laparotomy.
Sclafani  initially reported splenic artery infusion of pitressin, gelfoam pledget embolization, and coil occlusion of the proximal splenic artery as three methods of transcatheter hemostasis for splenic injuries in 1981. Coil occlusion of the proximal splenic artery, which was permanent yet did not result in splenic infarction, was considered the best method of splenic hemostasis and has been used since 1982 at our institution.
We postulate that a "pulsating hematoma" of the spleen is represented on arteriography by evidence of active arterial extravasation. Coil occlusion decreases temporarily splenic artery blood flow and splenic pressure, thereby allowing time for the pulsating hematoma to clot and the spleen to heal. Splenic viability is preserved through the rich collateral circulation of the splenic artery from the left gastric artery, the gastroepiploic artery, the pancreatic branches of the celiac and mesenteric arteries, and the omentum. [5,6] With the maturation of splenic collaterals, blood flow returns to normal. This is usually accomplished within weeks.
The two alternatives to splenic artery coil occlusion originally described were considered unsatisfactory. Pitressin has temporary and unpredictable actions. Gelfoam embolization was also felt to be limited because it occludes splenic vessels distal to collateral circulation and results in splenic infarction. It is, therefore, not preferable because maximum splenic salvage is the purpose of therapy. Complete gelfoam embolization of the splenic circulation should be considered only as an urgent preoperative maneuver to deal with refractory shock that occurs in the angiography suite and to enable patient transfer to the operating room. We have not been forced to perform this, although the use of a limited number of pledgets to control extrasplenic hemorrhage has already been mentioned in Results.
A review of 44 patients with CT-diagnosed splenic injury managed by bed rest or embolization and bed rest was published in 1991.  This study demonstrated that the absence of arteriographic extravasation from the spleen is highly predictive of successful outcome of nonoperative management. It demonstrated further that coil occlusion of the splenic artery effectively stops splenic hemorrhage and prevents delayed splenic rupture and increases the number of patients, including those with hilar injuries, who can be treated without operative exploration.
To our knowledge, transcatheter coil occlusion of the proximal splenic artery has not been reported in the surgical or trauma literature. This present paper, with a much larger patient sample, confirms the previously reported conclusions.
Splenic artery occlusion by coil embolization mimics the surgical technique of splenic artery ligation that has been successfully used to control splenic hemorrhage and allow splenic salvage by several authors. [8-11] Both maneuvers decrease temporarily splenic blood flow and arterial pressure by occluding the main arterial conduit to the spleen. The rich network of collateral circulation from the left gastric artery to the short gastric branches of the distal splenic artery, from the dorsal pancreatic artery and the pancreaticoduodenal arcades to the transverse pancreatic, and the pancreatica magna branches of the distal splenic artery, as well as the omental and gastroepiploic collaterals, all provide alternative blood flow to the spleen. [5,6] Viability of the spleen is maintained during the healing process. This explains the absence of splenic infarction!
after splenic artery embolization.
We believe that proximal inflow occlusion is compatible with maintenance of splenic immune function. Schwalke et al.  evaluated clearance of opsonized autologous red blood cells in normal controls and in 5 of 21 patients who underwent splenic artery ligation as part of salvage procedures for splenic trauma. They noted no significant difference in clearance between the two groups. Greco and Alvarez  have shown that the spleen undergoes hypertrophy, and as much as 80% can be removed. Even when supplied only by short gastric collateral circulation, it can still protect against a pneumococcal challenge.
Scintigraphy performed during convalescence in many of our patients has demonstrated that the reticuloendothelial system remains viable. Although we attempted to perform other studies of immune function (such as circulating tuftsin levels, IgM, and T-cell function), the nature of our patient population made this an impossible task.
The advantages of coil occlusion of the splenic artery are several. Mobilization of the injured spleen during celiotomy is avoided. This maneuver may exacerbate or initiate bleeding that has stopped spontaneously. In the one patient in our series who underwent laparotomy to treat a gastric herniation through a diaphragmatic rupture after coil occlusion of the splenic artery for a grade IV hilar injury, the spleen was left in situ and not mobilized. It was the opinion of the operating surgeon (T.M.S.) that attempts at splenorrhaphy would likely have failed. This is the only patient with operative correlation.
Our success with coil occlusion in treating bleeding grade IV splenic hilar injuries (84% salvage) extends the types and number of splenic injuries that can be managed nonoperatively. Most surgical authors assert that high-grade splenic injuries should be treated by prompt exploratory laparotomy. [14-18] These authors, however, were unable to save as many grade IV injuries as reported in the current series. In Cogbill's review, only 2 of 112 patients treated nonoperatively sustained grade IV injuries. Both patients failed nonoperative therapy, and neither spleen was salvaged.
Even when exploration is required, coil occlusion may facilitate splenorrhaphy just as splenic artery ligation has been recommended by some surgeons as an adjunct to splenorrhaphy. The three patients who had laparotomy after coil occlusion because of surgeon's preference, ultimately had splenectomy.
Coil occlusion is technically simple. The majority of these patients are otherwise healthy young adults in whom catherization and contrast medium pose little risk. Splenic artery coil occlusion requires selective catheterization only of the proximal 2 to 3 cm of the splenic artery, well within the abilities of most interventional radiologists. Coil occlusion, although not without risks, is simple: a coil is pushed through the catheter in front of a guidewire. No superselective catheterization, coaxial microcatheterization, or other difficult maneuvers are necessary.
Coil sizing, however, is critical. Coils too large may protrude into the celiac artery, and coils too small may embolize into the splenic hilum and block most collateral networks. The interventional radiologist, therefore, should be skilled in intravascular retrieval techniques to deal with this uncommon problem.
The present study also shows that arteriography still has considerable value as a diagnostic test even if the technical expertise necessary to perform coil occlusion were not available. The absence of contrast extravasation on celiac arteriography has reliably predicted successful nonoperative management in every one of our patients for more than 10 years. We have used this angiographic assessment to triage patients with CT-diagnosed splenic injury of all grades to bed rest and observation as primary treatment. None of these patient has required laparotomy for continued blood loss.
This answers a major deficiency of CT (i.e., its inability to predict outcome). Although CT has been shown to demonstrate splenic injuries well,  CT grading systems do not correlate well with operative findings. [16,19-22] Nor does the CT appearance predict which patients will successfully complete conservative therapy. [22-24] This has led to recommendations that grade III and grade IV injuries should be routinely explored. [15,16] Moreover, a CT scan is a single static image of the spleen. It rarely shows active arterial hemorrhage. It cannot tell whether bleeding is arterial or venous, or stopped or ongoing.
A number of CT grading systems have been proposed, but all have limitations as triage tools for determining which method of treatment should be followed. Resciniti et al.  described a system that was based on the extent of parenchymal and capsular damage, and the degree of hemoperitoneum present. The authors noted that a score below 2.5 was a reliable indicator of successful outcome of nonoperative therapy. Only minor injuries, however, result in such a low score, and following the authors' recommendations limits significantly the number of patients safely manageable by bed rest.
Buntain et al.  described a splenic injury grading system based on CT appearance. Grade I injuries resulted in pericapsular or subcapsular hematomas without splenic fractures. Peripheral splenic hematomas or fractures that did not involve the hilar area were described as grade II. Grade III injuries involved the hilum. Grade IV injuries resulted in fragmentation or devascularization. The authors recommended that some grade I and moderate grade II injuries could be considered for nonoperative therapy. They recommended exploration of all severe grade II and all grade III and IV injuries. This approach restricts the number of patients who can be managed conservatively. Nonoperative management was attempted in only 16 of 46 patients with CT-diagnosed splenic injuries in Buntain's series. Indeed, it is our opinion that nonoperative management of hilar injuries may result in higher rates of splenic!
salvage by obviating mobilization of the spleen.
Other reports--particularly those of Kohn et al.,  Becker et al.,  Mirvis et al.,  and Umlas and Cronan --have shown that the appearance of the spleen on CT does not correlate well with the ability to manage patients by bed rest. Some severe injuries can be managed successfully, whereas some lower grade injuries may fail. In addition, the ability to judge pathology and extent of injury by CT is unclear. Although Malangoni et al.  found that the CT grading commonly underestimate the operative findings, Becker et al.  and Mirvis et al.  have stated that the CT appearance could either underestimate or overestimate the extent of damage seen at!
Based on our results, however, CT may be sufficient and angiography may not be warranted in those patients with grade I injuries and no clinical evidence of persistent hemorrhage. Arterial extravasation was seen in only 3 off 28 patients with grade I injuries (11%). Perhaps it would be cost-effective to observe patients with such injuries without angiography until clinical conditions warranted it. Angiography may be limited to those patients with grade I injuries who have persistent blood loss or require blood transfusion.
On superficial review of our data, it seems that angiography may not add any benefit to the management of patients treated nonoperatively because the rate of successful nonoperative therapy in our series seems to be similar to that reported for nonoperative therapy alone, based on clinical assessment or CT appearance.
It is noteworthy, however, that the current study seems to increase the number of patients who are considered for nonoperative therapy. Eighty-eight percent of the patients who underwent CT examination were initially treated nonoperatively. Moreover, half of those who were initially explored were managed in that manner based on the preferences and prejudices of the traumatologist rather than objective clinical criteria.
This expanded approach to nonoperative therapy compares favorably to other reviews of patients with CT-diagnosed splenic injuries that have limited the number of patients based on the age of the patient, the clinical presentation of the patient, the CT appearance of the splenic injury, or the degree of hemoperitoneum present on CT. Moreover, most of these studies exclude patients who have had hemoperitoneum based on DPL. Smith et al.  reported a successful nonoperative management rate of 93% without the use of angiography or embolotherapy. Only 46 of 112 patients with blunt splenic injury, however, were actually managed nonoperatively, and only 33 patients were older than 14 years. 
Umlas and Cronan  reported a success rate of 92% among 43 splenic injuries diagnosed by CT and treated initially by observation. They did not attempt, however, conservative therapy in 13 patients. Moreover, 25 patients were children who universally did well. Analysis of their adult cohort demonstrated that only 15 of 27 adults were successfully managed by nonoperative techniques.
The applicability of interventional radiology for splenic injury in hemodynamically unstable patient remains to be determined. At this time, we do not have a clear sense of the circumstances in which coil embolization may have value in the unstable patient. For the present, it is our recommendation that all hemodynamically unstable patients be treated in the operating room. Intraoperative angiography and embolization may have uses in the future, but this requires further evaluation.
There are limitations to the design of this study. It was not a randomized controlled study, comparing bed rest alone, splenic artery occlusion, and bed rest and laparotomy. The primary purpose of this study was to determine the feasibility of using a nonoperative alternative to splenic artery ligation to treat splenic injury. In addition, blunt splenic trauma is not that common at our facility, and randomization was not practical. We compiled these 172 patients over 12 years. A collaborative multi-institution study is needed to obtain sufficient numbers to test the practicality and reproducibility of the results.
We believe that clinical assessment should be the primary determinant of the diagnostic and therapeutic algorithms used to manage patients with abdominal trauma. Those hemodynamically unstable patients who cannot be stabilized and who are suspected of intra-abdominal injury should be treated by urgent operative intervention. In hemodynamically unstable patients with equivocal findings, DPL should be used to determine hemoperitoneum. When DPL is positive in unstable patients, urgent laparotomy should also be performed. If DPL is negative in unstable patients, other causes of hypovolemic shock, such as hemorrhage from pelvic fractures, should be considered. In stable patients, CT provides more information and should be used when readily available.
It is more difficult to make a recommendation regarding severely head-injured patients and others whose clinical examination is more difficult. In some cases, CT can be used to exclude associated bowel and pancreatic injury, as well as neurosurgically treatable intracranial injury; however, if CT is at all equivocal, it is probably safer to perform laparotomy than anguish over the decision.
In summary, the CT appearance of splenic injuries cannot determine treatment because it is a static view of a dynamic process. We have shown that angiography can refine the characterization of CT-diagnosed splenic injury by assessing vascularity and hemorrhage. When angiography shows no arterial extravasation, nonoperative management is consistently successful. When used with transcatheter coil occlusion of the proximal splenic artery, angiography can increase the number of patients who are candidates for this management.
Dr. Steven B. Johnson (Lexington, Kentucky): This study represents the continuing noteworthy efforts of Dr. Sclafani and his associates at Kings County Hospital to expand the role of diagnostic and interventional angiography in the management of trauma. With this current effort, they have reviewed their 12-year experience utilizing angiography for the management of hemodynamically stable splenic injuries diagnosed by CT scan. The purpose of angiography was two-fold: to assess for dye extravasation from the injured spleen as an indication of ongoing hemorrhage, and to evaluate if splenic artery coil embolization was effective at controlling splenic hemorrhage among those patients with dye extravasation. They have clearly demonstrated that angiography can be performed in these patients, that angiography can demonstrate extravasation, and that splenic artery coil embolization can be accomplished without significant morbidity.
The authors should be commended for maintaining the protocol over such a long time period, especially in light of the significant changes in splenic injury management that have occurred over the past 15 years. Of the 172 hemodynamically stable patients entered into the study, 150 (87%) were initially managed nonoperatively according to protocol. Splenic injury grade was not a contraindication to nonoperative management. These 150 patients comprise the study population that underwent celiac and mesenteric artery angiography to evaluate for splenic dye extravasation. Ninety patients did not have any evidence of hemorrhage at the time of angiography, and 87 patients were successfully managed with bed rest alone. The remaining 60 patients had dye extravasation that was managed by splenic artery coil embolization. These patients had a 5% failure rate that resulted in splenectomy. The overall nonoperative success rate for this study was an !
impressive 98%. My concern is for the non-bleeding patients that underwent angiography caused by the inherent costs, morbidity, and mortality of angiography. This is especially important because the patients received two contrast dye injections under this management protocol. Considering that 67% of patients were not bleeding at the time of angiography, have the authors identified more accurate criteria to select the patients that could potentially benefit from this modality?
Regarding the use of splenic artery embolization, the study fails to provide sufficient rationale as to the benefits of this intervention. Unfortunately, there was not a control group that did not receive embolization. Without knowing the outcome of a concurrent control group, any implied benefit of splenic artery embolization is speculative. Their 5% postembolization splenectomy rate is identical to the 4.5% splenectomy rate noted by Cogbill's multicenter study on nonoperative management of splenic injuries. It would seem that embolization does not reduce the risk that a hemodynamically stable patient might subsequently require splenectomy. Can the authors explain how this intervention will benefit patients if failure rates resulting in splenectomy are identical to other studies not utilizing the procedure?
Splenic artery coil embolization, like proximal splenic artery ligation, should not be expected to maintain normal splenic immunologic function. Splenic scintigraphy can demonstrate perfusion, but does not accurately evaluate immunologic function. Studies evaluating splenic immunologic function after splenic artery ligation have demonstrated a marked reduction in function. Horton noted splenic artery ligation preserved splenic tissue, but that splenic function as assessed by pneumococcal clearance was depressed to an extent similar to splenectomy. Subsequently, Scher demonstrated that techniques that preserve splenic hilar blood supply confer the most benefit against pneumococcal bacteremia among various techniques for splenic preservation. Preservation of splenic mass and normal splenic hilar blood flow are important for maintaining immunologic function and clearance of encapsulated bacteria. Do the authors have any data regarding th!
e subsequent immunologic function of patients who have undergone coil embolization? Indirect evaluation of splenic function by red blood cell appearance on peripheral blood smears, although not as important, might provide evidence of residual splenic function. Can the authors comment further on the three embolized patients that developed sepsis? Do the authors consider these patients to have had "postsplenectomy sepsis"?
Regarding the technical aspects of this procedure, can the authors comment on their approach to injuries in different splenic locations? Are superior pole injuries more difficult to manage, because of the short gastric vessels, than inferior pole injuries? Does the presence of an aberrant polar artery originating from the celiac axis change how the authors manage these patients?
The authors should be commended for this interesting study and their fine presentation. With further study, this technique should allow future improvements in the management of splenic injuries.
Dr. Burton Harris (Boston, Massachusetts): Dr. Sclafani and his colleagues, who are third-generation Kings County Hospital trauma surgeons, work in the shadow of Gerry Shaftan whose teaching is to do no more than necessary. Despite that, they embolized every patient who was still bleeding at the time of CT. Is it a surprise that the spleens of recently injured patients might still be bleeding? If bleeding was the only discriminator, as it seems, this would have been a far more compelling study had those patients been selected out and then randomized for embolization versus nonembolization. Are you contemplating such a study? I congratulate the authors on this interesting, but purely preliminary paper.
Dr. Eric Frykberg (Jacksonville, Florida): I would like to congratulate the authors, too. It is a very interesting and thought-provoking study. I would agree with a lot of the limitations that have been brought already. What I would like to ask is have you all ever given any thought to using this in the unstable patient where you really could make a real contribution and alter management.
Dr. F. Barry Knotts (Toledo, Ohio): We have always used as a gauge of success of nonoperative management or operative splenic preservation the analysis of the number of units of blood utilized (i.e., how much additional blood loss was incurred by choosing the alternative to splenectomy?). I wonder if the authors have made any analysis of the relative amount of blood loss or their blood utilization during these procedures.
Dr. William Pfeifer (Erie, Pennsylvania): You comment there that you have several patients that had either abscesses or were septic and, at least to the larger question of those treated by your particular methodology herein, did you look at those that had secondary injuries or unfound primary injuries that later resulted in either surgery or complications and what that effect has on your cost impact?
Dr. Michael Muggia (Jerusalem, Israel): First of all, I envy the authors for their courage. I want to re-emphasize the point of blood infusions, because blood infusions probably compromise the immune system, could you comment on that? Second, I do not understand the whole issue. What is the point in keeping a ruptured spleen in an adult patient, as recent evidence suggests that major complications (early or late) after splenectomy are rare in adult patients.
Dr. David Feliciano (Atlanta, Georgia): How many of these patients had extrasplenic extravasation of the original CT? Second, if you had eliminated the patients with intraparenchymal hemorrhage only, what would that have done to your need for splenic arteriography in the study?
Dr. Gene Cayten (Bronx, New York): I noticed that you did not use any statistical analysis in comparing the three groups, and I was wondering whether you felt that your limitations to your study and your lack of randomization really precluded you from making any definitive statement regarding differences between those three groups?
Dr. Salvatore Sclafani (closing): I would like to clarify a couple of things first. When I said extravasation, these are extravasations seen on arteriography. We rarely see bleeding on a CT scan. I agree with you, David. I think that the first thing that we notice is that the grade I injuries tend not to have a positive arteriogram. I think that we are pretty much on the verge of seriously not doing arteriography in the grade I injuries. I must say that much of this came up because we had a grade I injury that we managed nonoperatively, and the patient went on to have a recurrent hemorrhage. We brought him back and did not do an arteriogram. He showed arterial extravasation within the organ, and we felt that we were dealing with a pulsating hematoma of the spleen. That is why we started doing this in the first place. There are a few patients with low-grade splenic injuries that we have over the time elected not to do arteriography, or!
the patient refused arteriography and it has come back to haunt us about half the time. I believe that when we do arteriography in those patients late, they are having arterial extravasation within the organ. I think that it is a predictor; it is very hard to prove. Our patient population is such that it is very difficult to do a randomized study. It would take another 10 years to do another randomized study to get 60 patients. I would like to do it with some other groups if I could get some interest in expanding this. I think it is worth looking at. I do not think that our institution is capable of doing a randomized study.
The patients who developed sepsis had complications of coiling. One of them had a coil that went into the hilum, and we could not retrieve it. There was, therefore, no blood flow beyond the collateral circulation; the organ infracted and got infected. The other patient developed pneumonia and also developed perisplenic infection with Staphylococcus epidermis. The organ itself was not infracted. There were a couple of small infarcts, but by and large the organ was viable at the time it was removed. It is hard for me to say whether the patient's immunity is really compromised. There have been studies that show that the perfusion of these organs with time is equal to the perfusion of, in other words, the flow within the splenic artery into the hilum is equal to a normal spleen after several months. This literature is largely in those patients who had splenic artery embolization for hypersplenism, and the hypersplenism is correctable only!
transiently. It is a difficult question, and we did try to get the patients for immunologic studies, but we just could not get them to come back to the hospital. I drove to their houses, and it was still unsuccessful. We tried to pay them to come to the hospital, but they would not come. It is a serious limitation, and I am not trying to make light of it. It is a real limitation of the methodology. I am not aware of any good studies that show that splenorraphy maintains splenic function.
The transfusion requirement among those patients who were treated nonoperatively by bed rest alone was 1.3 or 1.6 U. It was similar to those treated by embolization. Among the patients who were explored, it was just about double that. There is a decrease in transfusion requirement and two-thirds of the patients required no transfusions. The length of stay in the hospital is about the same as splenectomy, although we kept the patients in the hospital for up to 40 days when we first started using this technique, because we really did not know if it was going to work and we did not trust it. We keep them in the hospital about 5 to 7 days now, so it really does not increase the length of stay.
1. Fisher RP, Beverlin BC, Engrave LH, et al: Diagnostic peritoneal lavage. Fourteen years and 2586 patients later. Am J Surg 136:701, 1979 [Context Link]
2. Jeffrey RB, Laing FC, Federle MP, et al: Computed tomography of splenic trauma. Radiology 141:720, 1981 [Context Link]
3. Baron BJ, Scalea TM, Sclafani SJA, et al: Non-operative management of blunt abdominal trauma. The role of sequential diagnostic peritoneal lavage, computed tomography, and angiography. Ann Emerg Med 22:1556, 1993 Bibliographic Links [Context Link]
4. Sclafani SJA: The use of angiographic hemostasis in salvage of the injured spleen. Radiology 141:645, 1981 Bibliographic Links [Context Link]
5. Anderson JH, VuBan A, Wallace S, et al: Transcatheter splenic arterial occlusion: An experimental study in dogs. Radiology 125:95, 1977 Bibliographic Links [Context Link]
6. Yoshioka H, Juroda C, Hori S, et al: Splenic embolization for hypersplenism using steel coils. Am J Radiol 144:1269, 1985 Bibliographic Links [Context Link]
7. Sclafani SJA, Weisberg A, Scalea TM: Blunt splenic injuries: Nonsurgical treatment with CT, arteriography, and transcatheter arterial embolization of the splenic artery. Radiology 181:189, 1991 Bibliographic Links [Context Link]
8. Keramidas DC: The ligation of the splenic artery in the treatment of traumatic rupture of the spleen. Surgery 85:530 1979 Bibliographic Links [Context Link]
9. Conti S: Splenic artery ligation for trauma. Am J Surg 140:444, 1986 [Context Link]
10. Andersson R, Gustavsson T, Alwark A: Splenic artery ligation for traumatic rupture of the spleen: A case report. Acta Chir Scand 151:709, 1985 Bibliographic Links [Context Link]
11. Hadley GP: Splenic artery ligation: An adjunct to splenorrhaphy in children--Case reports. S Afr Med J 66:578, 1984 Bibliographic Links [Context Link]
12. Schwalke MA, Crowley JP, Spencer P, et al: Splenic artery ligation for splenic salvage: Clinical experience and immune function. J Trauma 31:385, 1991 Bibliographic Links [Context Link]
13. Alvarez FE, Greco RS: Regeneration of the spleen after ectopic implantation and partial splenectomy. Arch Surg 115:772, 1980 Bibliographic Links [Context Link]
14. Cogbill TH, Moore EE, Jurkovich GJ, et al: Nonoperative management of blunt splenic trauma: A multicenter experience. J Trauma 29:1312, 1989 Bibliographic Links [Context Link]
15. Buntain WL, Gould HR, Maull KI: Predictability of splenic salvage by computed tomography. J Trauma 28:24, 1988 Bibliographic Links [Context Link]
16. Pachter HL, Spencer FC, Hofstetter SR, et al: Experience with selective operative and nonoperative treatment of splenic injuries in 193 patients. Ann Surg 211:583, 1990 Bibliographic Links [Context Link]
17. Resciniti A, Fink MP, Raptopoulos V, et al: Nonoperative treatment of adult splenic trauma: Development of a computed tomographic scoring system that detects appropriate candidates for expectant management. J Trauma 28:828, 1988 Bibliographic Links [Context Link]
18. Smith JS Jr, Wengrovitz MA, DeLong BS: Prospective validation of criteria, including age, for safe, nonsurgical management of the ruptured spleen. J Trauma 33:363, 1992 Bibliographic Links [Context Link]
19. Buckman RF Jr, Dunham CM, Kerr TM, et al: Trend toward nonoperative management of splenic injuries. Surg Gynecol Obstet 169:206, 1989 Bibliographic Links [Context Link]
20. Malangoni MA, Cue JI, Fallat ME, et al: Evaluation of splenic injury by computed tomography and its impact on treatment. Ann Surg 211:592, 1990 Bibliographic Links [Context Link]
21. Kohn JS, Clark DE, Isler RJ, et al: Is computed tomographic grading of splenic injury useful in the nonsurgical management of blunt trauma? J Trauma 36:385, 1994 Bibliographic Links [Context Link]
22. Becker CD, Spring P, Glattli A, et al: Blunt splenic trauma in adults: Can CT findings be used to determine the need for surgery? Am J Roentgenol 162:342, 1994 [Context Link]
23. Mirvis SE, Whitley NO, Gens DR: Blunt splenic trauma in adults: CT-based classification and correlation with prognosis and treatment. Radiology 171:33, 1989 Bibliographic Links [Context Link]
24. Umlas S, Cronan: Splenic trauma: Can CT grading systems enable prediction of successful nonsurgical treatment? Radiology 178:481, 1991 [Context Link]
Accession Number: 00005373-199511000-00004